The 60Co Calibration of the ZEUS Calorimetersdigitool.library.mcgill.ca/thesisfile61196.pdf · The 60Co Calibration of the ZEUS Calorimeters by Ling ... 3.1 Radzation Dose from a

The 60Co Calibration of the ZEUS Calorimeters

by

Ling-Wai Hung Department of Physics, McGili University

Montréal, Québec July 1991

A Thesis submitted to the Faculty of Graduate Studies and Research

in partial fulfillnlent of the requirements for the degree of 11aster of Science

@ Ling-Wai Hung, 1991

( ...

Abstract

The calibration and quality control measurements of the ZEUS forward and rear calorimcters made using movable 60Co sources are described. The types of assembly faults discovcred from the l'uns taken first at CERN and then later at DESY from mid 1990 ta carly 1991 are prcsented. The VILS attenuation length as weIl as changes in the attC'lluation length of the scintillators can be monitored by the cobalt scans.

.. Il

-

Résumé

Le processus de calibration et de mise au point des calorinlC'trcs avant ct arriht> du détecteur ZEUS en utilisant ulle source mobile de 6OCO est d('C1it en lktail. L('s différentes erreurs d'assemblage découvertes lors des (,Xpl-I iell(,(~s rtH CEHN pHI!> à DESY au cours des années 1990 et 1991 sont égal{,!llent présl·ntées. L'clttéllll.\tion d('s dephaseurs et les changerneI1ts de la longueur d'atténuation d('~ ~cilltilli\klll" peu\'('lIt

être estimés en utilisant une source de 60Co.

iii

l

Contents

Abstract ii

Résumé Hi

List of Figures vi

List of Tables ... Vill

Acknowledgements ix

1 Introduction 1 1.1 Background 1 1.2 Kinematics 3 1.3 HERA Physics 6 1.4 The ZEUS Detector 8

2 The Calorimeter 14 2.1 Calorimetry . •• 1 , •••••• 14

2.1.1 Sampling Calorimeters . 15 2.1.2 Eledromagnetic Showers 16 2.1.3 Sampling fraction . 19 2.1.4 Hadronic Showers . 20 2.1.5 Compensation ... 22

2.2 The ZEUS Calorimeter . . 24 2.3 FCAL/RCAL Modules ... 25

2.3.1 DU jscintillator stack 29 2.3.2 Optical Readout .. 32 2.3.3 Straps . . . ....... 34

3 60CO Source Scanlling 35 3.1 Physic3 of Light Production 37 3.2 Inside Scanning . 38 3.3 Outside Scanning ..... 39

1 3.4 The Source Wire 40 ~ 3.5 Safety 42 ......

iv

n.

CONTENTS v

3.6 Preparation · . 4~J 3.7 Scan sequpnce . -13 3.8 Thf' Driver. · . H 1.9 The outside driver 4i 3.10 Data Acquisition 4ï

4 '1onte Carlo 55

5 R,· 'ults of €O°Co Scans 65 .5.1 60C 0 Signal . . 65 5.2 Rcproducibihty .. 69 5.3 Faults ... · . . . 71

5.3.1 Bad Stacking il 5.3.2 Bent WLSjOpticai Contact 73 5.3.3 WLS problcr, .. . . . 75 5.3.4 Light Path ( t,rucHons 77 5.3.5 InhomogeneitJt?s .. , •• 1 79

5.4 Scintillator Attenuation Lcngths . 80 5.5 WLS Attenuation .... .... 82

6 Conclusions and Outlook 84 6.1 Scanning in the ZEUS Hall . 84 6.2 Conclusions ......... 85

References 88

...

(

List of Figures

1.1 Layout of the HERA Rmg . . . . . . . . . . . . . . . . . . . . . . .. 3 1.2 Feynman dwgrams of lowest order a) Ne and b) CC interactIOns .. 4 1.3 Polar dwgram (lf the kinematics for the final lepton (upprr parI) and the

CUITent Jet (lower part) ll'lth Imes of constant :r and Q2. Connectmg a gwen (:r, Q2) PO/lit wlth the 01'lgzn gives the laboratory momentum vutors, as shown by the e:ramplc for x = 0.5, Q2 = 5000Ce V 2. 7

1.4 Vicw of the ZHUS DetectaI' ... . . . . . . 9 1.5 Cross SectlOnal VICW of the ZEUS Detcclor. . . . . . . . . . . 10

2.1 FractlOnal Encrgy Loss for Electrons and PosItrons in Lead (j7-om Rev. of Pari. P1'Op., Phys Lei. B. Vol 239) . . . . . . . . . . . . . . . . .. 17

2.2 Pltoton C7'oss-sectlOn ln 1/ ad as a functlOn of photon cnergy. (from Rev. of Pari. Prop. 1980 ((fillOn) . . . . . . . . . . . . . . . . . . . . . .. 17

2.3 Average deposlted entrgy vs dcpth for electromagnetic showcrs of en-ergies 1. 20 and 75 Ge V . . . 19

2.4 Posttioning of FeAL modules 28 2 . .5 Posttioning of ReAL modules 28 2.6 View of a Large FeAL Module. 30 2.7 Thlckness D,stribution of FCAL seintillators . 31 2.8 Tyvek Paper Pattern used for FCA L EMC Seintlilators 32 2.9 View of WLS reading out scintillator . . . . . . . . . . 33

3.1 Absorptwn and Emission Spectra of SCSN, Y-1, and PM cathode 49 3.2 View of Ihe Source GUlde Tubes and Light Guides 50 3.3 Skelch of the outside driver scanning a module. 51 3.4 Source wire used in inside scannmg . . 52 3.5 Source wtre used zn outslde scanning 52 3.6 Layout of Elements in the Run Control 53 3.7 View of Source D/'wer from above 54 3.8 Vlew of Sour'cc Drwcr from the side 54

4.1 Layer' Structure used in Monte Carlo Simulation. Dislallrf'(I are in mm. 58 4.2 Monte Carlo of the Response function, R(z), of a Singl, '"', lIt/lllator,

before and after smoothmg . . . . . . 59 4.3 Real Response of a Single Scintillalor 59

VI

-l

LIST OF FIGURES VII

·1.4 lOTleoluled Response Function Repre~(nll1!g SlgTlal F/'om /lAC. Cll ·1..5 .\fol/te Codo flAC sectlOll Il,/th OIH' sem/tllalor 50f//> ... hlldoll'u! 61 4.6 .llonte Carlo of HAC Sfrl/Oll Il l ltl! dll1Jwgcd HL<; l'hf ji/'8/ .3:1 IOyf l'.'

haue a rcdlldlOn III lzghl output of :!09é fo/' thf jll',.,1 layf /', dl('1"u],'II1!1

111IW1'ly to 0% fol' tlu tIN nly·jirst la y(/', , t;2 -\.7 ,\fonte Carlo of /JAC sfcllOlI 1L,/th !!() layers jJlL • .,1!U{1II by (,')llll ()',!

4.8 Manie Car/a of /lAC ser/101l u'dh 20 layfrs Pll,-lud ouf (, 111111 {\:3 4.9 X- y dlstrzbutlOn of entrgy dfpOSlffd ln sfllltlllator ll'dh the .""Ol/J'('l ]l0-

siflOncd at 8 mm awoy, as !Il oufsll!c :,calll11llg. (l·t

5.1 Raw slgna/ from PAls n:adzng out bath s1de8 of Il 114(' ... ('cito1l G6 5.2 SIgnais from PAIs rtadmg Ollt both suies of a n0/'11Iai /lAC scellOn 67 .5.3 Signais fl'o»1 P.'rls rmdwg ouf bolh sliles of a normal HM(' ~I('IIOII III

FeAL 67 5..1 Signais from P'\[s rcadmg out both sides of a normal Il..1 co ~tcflOlI 111

FeAL . . 68 5.5 SIgnaIs fl'om PAIs reading out both s!df.S of a normal FU(' ln ue'AL. G8 5.6 Scans of Module CDN2, Tou'cr 14 llAC2 u'dh the l/I,.,u!r and oul ... u{c

methods 69 .5.i n'ldth of HACt SfctlOTlS from inslIlc and ouls/dr 8('(11/11/119 70 .5.S Bad sfacking . 72 ,5.9 S}lifhd Scwlzllmor ï:3 5.10 lncrrased Rfsponse from Shlflcd Scwllllaior ln NL14 7·1 5.11 Sli/fied EMC WLS. Nole the shal'p drop where the fir!'t sCIIl/Illa/or [(Jillr

should be md/catl1lg fhat the first layer lB not seen 76 5.12 Sttcking back 1'efiector. 77 5.13 Shadowed Scwttllator . 78 5.14 Bad homogenClty .. 80 5.15 Dl~tnbutlOn of the quantlty e-d/ À ln FeAL [{ACl II/es 81 5.16 Cen:nkov light scen ln a source scan wtth the semtilla/ors eovered 82 5.17 Attenuation lengths of EMC WLS . . . 83

List of Tables

1.1 Paramders of the liERA collider . , ......... 4

2.1 ComposltlOT! of a Samphng Layer in EMC and HAC. 26 2.2 SU1/lmary of the dl11unslOns of FeAL modules .... 2ï 2.3 Summary of thc dmUIlSlOns of ReAL modules 27 3.1 Radzation Dose from a 2 mGz 60Co .source vs distance. 42 6.1 Summary of Faults in FCAL . 86 6.2 Summary of Fau/ts ln RCAL . 87

(

VllI

-

:\CI\NOWLEDGEMENTS IX

My work would not be possible without tht:' cOlltributioll" by tl\(' IIH'mlwr" nf tht>

ZEUS collaboration \\ho dcslgned. huilt and preparcd tht' ZEt1S ,'\IUIIII\t'tf'1 1 wlllllt!

likc to thank t}Wn1 for allowing n1(' to pal ti"lpak III ..,\tch (\ WOI t hwhilt· pJl)jt'ct. 1 woultl

also like to thank my sllpervisar David Hanna for al! th(' ad\'ICt' ,md Illotivatiull \-\"Ieh

he provided.

Life in Germany would not ha\'(' bccIl tll(' salllc' Wlt hout t hl' suppurt of th,· fl'I

low CanadJan students, post-docs and profc",solS WOI hing at DES)' who "'t'If' .t1w<lys

willing to help or offer ad vice. AmuIIg them, 1 woult! Mt' to part icularly t ballk J),tVl'

Gill\inson who clic! Illllcb of t}J(' bllildlng alld p\'()grilllllllillg of III<' "0111'('(' dli\t'I, mu:..t

of the insldc :-,canning and, pel h,tpS 1110r(' Îll1pOI tant Iy. illt lodll< <'d 111(' tn IIIp;ht hf(· in

lIamburg.

1 would like to gi\'c very specIal thanks \II\" Gt'!llIan coll('agm'''' CI'nit Cloth. Bodo

Krebs and Focko ~l('ycr, who COI1C.t ructf'd t h(' outsidC' ,,( é-llllwr .\Ild "1)('111 wit Il Ille /1 ltl Il)'

long ddyS and mghts t>canning t\)e modules C('rllt an 1 Fue ho "!lOw(·d /Il(' hu\\" !o 11<;('

their analysib programs as weil pro\ Ide<! me wlth tl)(' \VLS and ''l'ill! illettor attl'llll,tllOIJ

data. Bodo, who led our }ittle group, did llluch of the \ i~\)a! ,lIlaly"!''' of tllC' '>Olll'c('

plots, drew the source driver diagrams and offered IIIally hl'lpf1\1 i>lIggestions.

And 1 must thank my friends and fami!y for their cont in1\al sllpport thlollghollt

this entire endea\"our.

_ ..

('

(

Ch.apter 1

Introduction

1.1 Background

Inquiring minds ha\ (' orteIl posed the questions; Who are we'? and What are we

made of? While phdosophers are still tackling the former, modern science has made

considerable plOglcss towards answering the latter, at l('ast 111 a malerial senSè, during

tll(' pa~t ct'nlury and a halL 5tarting with Mendeleev, who a little over a hundred

j'l'ars ago 01 ganized the PNiodic table, our insight into the underlying structure of

Nat me has grùwn increasingly sophisticated In his time dtoms were J<,lieved to

he the ~lllal1('~t nnits of matter. By 1932 the discO\enes of the electron, proton

and neut IOn :"('l'med to fully explain the structure of the periodic table and it was

thought that they ' ... ere the basic constituents of all things. However, later experiments

using cosmÎc rays and accekrators revealed a ri ch spectruffi of hadronic partic1es,

implying the existence of a great variety of fundamental particles. In 1964 Gell

Manil and Zweig proposed that the multiplicity of hadrons could be explained by

5U(3) S)'11l1l1<'tlY through the introduction of even more elernentary particles call~d

quarks[l]. Today we believe that ail matter is built from six quarks, six leptons and

thC'il' antipal t Ides. They interaet through the intel change of gauge bosons associated

with four fUlldamclltal forces, strong, weak, eledromagneti( and gravitational, which

themselvt's may be manifestations of a single unified force. Already the wcak and

1

CHAPTER 1. INTRODUCTIO.'l .)

elcctlOmagnetic forces are known to be asp{'cts of a single ('lt-ct ro-wl'ak forcI'. .\ Il

paIl ides and threc forces (gravitation excluded) art' acco\llllt'd for by the C\lllt'Ilt l~'

held theOlY callf'd the ·~tanoard model', ~'he glo\\'th uf Olll lllldn ... t'llldllig flUIll thl'

jJcllodic tdblc to 1 !te ... tillidard rno(kl cali lcugt'ly 1)(' c\tt1 dllltt'd 10 tlll' lapid ~IIC('I'~"'llIll

of experiments stlInulating or coufillning, theoretical ideas whit h III tmll forllH'c! tIlt'

bases for further cxpCrimE'llts,

One type of stlch cxperimeuts which has llliHlf' Ill1poltant contllh\ltinll" to tIlt'

de\'eloprnent of the standard rnodel, and in pttrtlCttlar to t he urHkr~talldlllg of 1 lu'

structure of the 111ld('on, is the scattcriltg of leptons by hadlOll" Sill({' ll'ptoll'" tll('

consickrcd to be point particles, they art' ideal for prohing UI\' su b..,t 1 IH t \lit' of t hl' 1111-

cleon. Evidence for the existence of poiutlike, charged 'pal tons' wltr.ill tlle proton Wd.S

first provided in 19G8 at SLAC where cl('ctrons of ('!lf'rgit's 9-16 Ce\' \V1'ft' sca.1t('I<'<! off

protons[2]. Tberc, the s, "th'ring CIOSS sect ions at Ilig'l mOIJWI1! \lm 1,1 tllI"fl"'" l}('h,I\'('d

more like the t'lastic scattering off point-like obj('(ts tlltlll lih(' the ilW],1stl< h!('dhllp

of the proton, Thcse pctrtons were c\'cntually idcntifieJ with tilt' 'lll.llk., pn'violhly

pO'itulated by Gcll-:\taon and Zweig to cxplalO the hadronic Sp<,ctllllll. lIadl Ull'> al'('

belie\'ed to be made of two or three vcl.lence quarks in il ~Cd. of \'11 tuaI quark-c\llt i'lUIU k

pairs. The quarks carry a colour charge and interact maillly tlllollgh th/' .,trollg fo!('('

mediated by the exchange of gl lions in analogy to the e!cd rOlll,lg!wt,ic fOI Ct' W Iw\('

the photon is the mcdiator.

The HErtA accelcrator (sec FigUl~ 1.1) at DESY in ILul1burg will colltd(' :W G('V

electrons with 820 Ge V protons yi('lding a centre of ma:',!> ((Jl\~) ('Ile! gy of v-;, -.: :H 1

GeV[3). As the world's first electron-proton collider, lt. will CO\'('f a rang!' of IIlO/lWn·

tum transfers (Q) reaching up to Q2 '" 105(GcV/c)2, a gn'iü 11I('[(',t,>e fl'Illn pr('\'ioll:',

experiments using fixcd targets which could only f('ach Q2 of il. few hlllld! ('(1 (nt v / r)2,

However the inteJaction cross section dccrcascs roughly as 1/Q1, rllakillg bigh Illmi

nosity a priority in order to acquire sufficient statJstics in the high Q2 1 egion. II EllA

therefore has 210 bunches of both electrons and proton') ('aeh con!.1111 i Iig 0(10 11 ) par

ticles. With a bunch crossing occurring every 96 ns, a luminosity (,( 1 ri' l03i CT/t-2 / S

CIl APTER 1. INTRODUCTION

\

\. ,.,.;~ v . • • \. / .... ~....)f/ ...

\ ..........

HER"

•••••

Figure 1.1: Layout of the HERA Ring

3

can he obtained (see Table 1.1). The distance scale orthe interactions hetween leptons

and quarks which is accessible by HERA is of the order 1O-18cm.

1.2 Kinematics

The principle processes in deep inelastic scattering at HERA energies occur through

neutral current (Ne) interactions exchanging a "( or ZO boson with a parton from the

proton, and charged current (CC) interactions exchanging a W:I: (see Figure 1.2). In

CC processes, the final state lepton is a neutrino and leaves the detector undetected.

In both cases the scattered parton is seen in the detector as a 'jet' of hadronic par

tic1cs emitted in a narrow cone balancing the lepton in transverse momentum. The

remnants of the proton continue on and most are 105t down the beam pipe.

There are only two independent variables in the overall event kinematics, both

of which can be measured Crom either the electron or jet (using the Jacquet-Blondel

CHAPTER 1. INTRODUCTION 4

proton eJectron Parameter ring ring units Nominal energy 820 30 GeV c.m. energy 314 GeV Q~QZ 98400 (GeV/c)1. Luminosity 1.5.1031 cm-1.,,-l

Number of interaction points 4 Crossing angle 0 mrad Circumference 6336 m Magnetic field 4.65 0.165 T Energy Range 300-820 10-33 GeV Injection energj 40 14 GeV Circulating currents 160 58 mA Total number of particles 2.1 . 1013 0.8.1013

Number of bunches 200 N umber of buckets 220 Time between crossings 96 n" Total RF power 1 13.2 MW Filling time 20 15 min.

Table 1.1: Para met ers of the HERA collider

e

e

y.z·

p p

q'

Figure 1.2: Feynman diagrams of lowest ortler a) NC and b) CC interactions .....

( ...


mcthod[4]) alone. 1'0 define the kincmatic relations we let Pe and PI be the four vectors

of the incollling and scattered lepton respect1\'ely, P that of the incoming proton, and

Cf = pf' - PI that of the exchange boson.

Tite total invariant mass squared is

The approximation assumes the particle masses to be zero, which is very reasonable

at lIERA energies. The square of the four momentum transfer, QZ, is defined to be

positive

Q2 - _q2

:::: -(Pe - PI)2

~ 4E E . 2 (JI e 1 sm 2'

The encrgy of the CUITent in the target rest frame, Il is

Il == p.q

mp

2Ep 201 ~ -(Ee - Eicos -).

nlp 2

The dimensionless variables Bjorken-x and y are given by

Q2

2P·q x =

y

Q2

2mpll

EeE, sin2 ~ ~

Ep(Ee - El cos2 ~)

-

=

::::

~

p.q p. Pe 2P 'q

S Il

IIma:r

Ee - EICOS2 ~

Ee

CHAPTER 1. INTRODUCTION fi

They are always in the range 0 $ (x,y) ~ 1. In the parton model, x can be identified

with the momentum fraction of the proton carried by the struck parton wllC'reag y

is a measure of the inelasticity of the event. A nothC'r uscful quant ity, the invariant

mass ~V of the hadronic system, is given by

w2 = (q + p)2

2 1 - x 2 = Q --+mp.

x

An accurate measurement of the kincmatics of an event requires a high n'solut iOIl

hermctic calorirneter. III the case of CC proccsscs, one cannot \11eaSl\I'e tht" o\ltgoillg

lepton and must instcad rely on only the jet meaS\lrCllwIIts for a dt'sn ipt ion of t.he

event. Even ",hen the clectron is present, as in Ne proceS~l'S, t.he ('If'ctron is oftf'n in a

kinematic reg,ion where a small error in the measure'ment of the t'le( tWIl'S /llOIlH'lltlllll

means a large error in x or Q2 (see Figure 1.3), Although traching detf'ctor~ aIso

give momentum information, their fractional resolution, <7p /p, grows linearly with

p. In addition, they are effective only for chargcd particle~. On the othel hanrl, a

calorimeter has a fractional resolution improving as liVE and is sensitive to same

neutral particles as well as charged ones. Using the calorimctf'r for Ilwasuring the

event kinematics also requires it to have fine segmentation for good angular n'solution

and good hadron-electron discrimination to distinguish electrons flOm pions.

1.3 HERA Physics

A broad range of physics prospects is available at HERA [5]. Structure functians,

h{~avy quark physics, compositeness, low-x physics and searches for exotic particles

such as leptoquarks are aH potcntial of areas of rescarch. III particular, the study of

the evolution of the structure functions of the proton with Q2 is of prime importance.

These structure functions, whià dcpend on the probing leptons' polarization, arc

related to the quark and gluon densities of the proton. The cro"" ""ri ion (or dcep

.'

l

("

GHAPTER 1. INTRODUCTION 7

FINAL LEPTON S 4

CURRENT JET 100 GcV t- of

Figure 1.3: Polar diagram of the kinematics for the jina/lepton (upper part) and the

current jet (lower part) with lines of constant x and Q2. Connecting a given (x, Q2)

pOint with the origm gives the laboratory momentum vectors, as shown 611 the example

for x = 0.5, Q2 = 50(lOGeV2•


inelastic scat tering is expressed in terms of the structure functions Ft, /<'2 and F3 by

d2a(e-p) lr.a2p2(Q2) 2 2 ' ,/2 ,) d.rdQ2 = r {y ,rFt(.r,Q ) + (1 - y)F'l(.l',Q·) + (y - '2 .rl-'J(J·,Q"))}.

The propagator P(Q2) is of the fùrm 1/Q2 for sillgk Î exchdngp, 1/(Q2 + ,\I~) fm

single zo exchange, 1/ Q2 + Mf~') for single \V± exchange and 1/ Q2 (Q2 + At ~) al'> the

Î - Z interference tenn.

The forms of the propagators show that th<, pure Î ('xchangc t('1'111 compld('ly

dominates at low Q2. HERA is however unique bccausc it can aUain high Q2 valul's

where the contributions from H:±, Z exchange becomc important.

The structure fundions are predicted by QCD (quantulIl cIlIolllooYllamin:) to

deCl'ease logarithmically \Vith increasing Q2 duc to gluon ernis~ion by the scat.tetcd

quark. Precise measmements of the structure functions at Il EHA can provide a

stringent test of QCD as well as search for new substructul'C 111 e!ect rons or quarks.

1.4 The ZEUS Detector

The ZEUS detector, as shown in Figures 1.4 and 1.5, is one of two dctectors that will

investigate e-p collisions in the HERA ring[6].

!ts development and const.ruction rcquired the cornbincd efforts of i'I ('Ollahoration

of 50 inst.itutes from 10 C'Ountries includîng about 350 physicists, Hs pl Îllctplc feature

is a high resolution fully compensating calorimeter. The asynlIllf'tJ'y of the lIERA

collisions, with the cms frame moving in the forward, or prot.on direction, is rcflccted

in the design of the detector where more emphasis is placed iu the forwélrd direction.

Here, an overview of the entire detector is presented.

Closest to the interaction region is a vertex dctector (YXD). II, lises a time ex

pansion drift type ceH to detect t.he decays of short lived particles. With an impact

parameter ] esolution of "'" 50jlffi it can detect particlcs with lifctillle~ greater than

10-1: s. Next comes the central tracking detedor (CTD). It is a cyllfldrical jet type

drift chamber with an outer radius of 85 cm and a length of 2·10 cm. There are 9

superlayers each with 8 layers of sense wires to measul'e track positions and dE/dx.


r

\

( Figure 1.4: View 0/ the ZEUS Detedor

•

-


Figure 1.5: Cross SectlOnal View of the ZEUS Detedor

Four of the superlayers have wires at a stereo angle of about 5 degrees in arder to

attain roughly equal angular resoJution in the z and azill1uthal directions. The pre

cision is belter than lOOJ./m. For particlcs going perpendicular ta the magnctic field

the expected momentum resolution is

u(p) - = 0.0002 . p $ 0.003, p

where pis in units of CeV le and ffi means addition in quadrature.

Complementing the CTD (Ire fOl'ward and rcar planar track detedors (FTO, RTO)

consisting of three and one planar drift chambers rcspectively. A transition radiation

detector (TRD) consisting of 4 modules sandwiched between the FTO chambers im

proves electron identification in the forward rcgion. Around these inner deledors is

a. thin superconducting solenoid providing a magnetic field of 1.8 T.

Surrounding the entire solenoidal coil and the trackint~ dctectors is the ZEUS

calorimeter. It is a high resolution sampling calorimeler made from layers of depleted

uranium (DU) plates interlea\'ed with plastic sdntillator tiles. The scintillator Iight

from two sides of a til'. is transported via wavelength shifter cars (WLS) and light

CliAPTER 1. INTRODUCTION 11

guides (LG) to photomultiplier tures (PM). Mechanically, the calorimeter is divided

into thrce main (()rnponents, th,.. forward (FCAL), barrel (BeAL) and rcar (RCAL)

(dlorill1ctels, \Vith a dcpth of 7,5,4 nuclear absorption lengths lespectively to match

the asymmetry of lIERA 's collisions. FCAL co vers the polar angles flOm 0 = 2.2°

to 39.9°, BeAI. f'xtcnds from 0 = 36.7° to 129.2°, and RCAL cO\'ers 0 = 128.1° to

176.5°. Togcthcr thcy coyer 99.8% of the soHd angle in the forward hemisphere and

99.5% in the backward hemisphcre. Longitudinally they are segmented lOto 2 parts.

The inner part, wlth a depth of about 25 radiation lengths Xo ('" one nuc\ear ab

sorption lcngth, .\), is called the electromagnetic calorimeter (E~C). The remaining,

outer part is callcd the hadronic calorimeter (HAC) which is further subdivided into

two subsections (HAC1 and HAC2) in the FCAL and BCAL components. The di

vision hetw('C'n the EMC and BAC sections greatly helps in distingUlshing between

clectromagoetic showcrs (è, Î) and hadl'onic showcl's because incident electrons or

gammas will deposit alm05t aIl their energy in the EMC section whereas hadrons will

deposit a subst.antial fraction in the HAC sections. The EMC and HAC ~ections in

FeAL and ReAL are nonprojective. For DeAL the EMC section if) projective in

both the polar angle, 0, and the azimuthal angle, 4>. The HAC sections are projective

only in <p. FeAL and RCAL arc divided into tower cells of 20 x 20 cm2 for readout

in the BAC sections. In FCAL and RC AL the EMC towers, except for those behind

DCAL (called HACO), are further segmented into 5 x 20 cm2 and 10 x 20 cm2 sections

rcspectively. To hclp dist inguish electrons from pions, especially within jets, provi

sions have becIl made to place two layers of 3 x 3 cm2 silicon detectors, called hadron

electron separators (lIES), at a depth of 3 and 6 Xo inside the EMC section of FCAL.

Re A Land BCAL have space for only one layer after 3 Xo. The small size of t.he RES

diodes and thcir forward p05itioning allùw them to differentiate between electrons

and hadrons because electron showers are narrowcr and develop more quickly than

hadron showers.

The cnergy resolution for the calorimeter determined from test beams is for elee

trous f7(E)/E == 18%/v'Eœ 1% and for hadron~ O'(E)/E == 35%/VF - ?%.

-


The backing calorimetcr (BAC) surrounds the high resolutioTl calorimt.'tt'r in or

der to measure the ellcrgy of late showcring particles. It uses lroll pl<llt's él" ah.,orbcr.

which also form the magnet \ ~oke, and plOj)ort ional tul)('s of alllmillllll1 as tht' sign.t1

layers. The resolutlOl1 for hadrons is a( E)/ E == 100;; / Ji:; B<'llIlld t!tt' ~'(l~(' al pt!$'

muon cletectors The banel and 1car muon delectors \I~(' lilllltt'<l -;h'('<lIJ}('1 ! IIh('~ to

identify lracks penetr,ümg tL:- calOlimeter, They Jdfc[<'Iltiatt' lW(\\'('('1I ('\(,Ilu, clll"lllg

from cosmic ray and halo muons background from t!1Ose oflglllafill~ III the illtt'l'ac

tion region. The fOl'ward muon spectromcter (F ~1l' 0:$ pfl)\'\dt'<' cUl i 1It!t'pt'lIdt'nt

momentum me!isuremcnt of muons as weil as illlpro\'ps t Iw It'j('c! ÎOII uf 111\1011'> fhml

background 50"ICCS, It uses a toroidally magrwtiscd Iton regie.1l illt(·d(·cl\\'d wlth dlift

chambers, limitcd 5tlcamer tubes, and time-of-fhght COllllt('r'3 A kadin~ proton sp('('

tlOmcter (LPS) tags elastically scattercd protOIlS which would 01111'\'\\'1">(' "-;(',q)(' tlOWII

the beam pipe, In addition, luminosity monitors upE>trealll (in t ht' proton dir('ct ion)

from the intera<.tion lf'gion detect photons and electrons from brermstrahlllng events

as wel! as flag low Q2 photoproduction events.

The detector is structurally divided such that inner compont'nts art" support(·cI hy

the bot tom yoke and outer ones are on a retractable clam shell.

In an effort to achie\'e and maintain ils design energy r(,~o!lIt ion, the ZEUS

calorimeter was built undcr strict quality control and cquipped with Vétl ious cali

bration systems. This thesis dca!s with the description of one suelt "y~t('m which US('S

movable 60Co sources to scan for assembly ercors as weIl as ta monitor th(' iong t('J'm

stability of the optical rcadout. In the following ('hrlpters the operai i(m and rpsu!t.s

from the 60Go scans first taken during the, summer of 1990 at CERN and thcll laler

at DESY from the faIl of 1990 to the spring of 1991 arc describcd. Chapt<'r 2 gives an

overview of calorimetry as weIl as a detailed descriplion of the caloIÏmcter. Chapter

3 provides details of the actual operation while scanning with GOGo sOllrces. Chap

ter 4 presents sorne results from Monte Carlo simulations of the cobalt source ~cans.

Chapter .) summarizes the findings from the initial cobalt. scans taken 1 ~fore the in

stallati0n of the calorimcter in the detector. Finally Chapter 6 gJVt>s :-,onw conclusions

CllAPTER 1. INTRODUCTION 13

( and a glance at the future applkations of the cobalt calibrat.ion system.

(

Chapter 2

The Calorimeter

2.1 Calorimetry

A calorimcter i~ a dcvice to mcasurc the cnergy of a parti< le f'utPIlllg it Il)' fully

absorbing its energy and producing a mcasurahle signal propol t iOllal 10 II. \\'})('II d

particle enters a calorimeter it will interact with the méttpridl, produCilIg s/'C(mc!ary

partidcs which in turn produce more particks propagatÎlIg tllloilgh lIl(' citlorillll'l«'r.

As this continues, a partiele 'showcr' develops. T\\'o types of shuw(', s (aIl J)(' dis

tinguished; electromagnetic and hadronic. Elcctromagnptic <.11' ,II l'r., .-lI't' made' lit> of

gammas, electrons and positrons which i,ltc"act only clcctf()llIaglH't I( ally through a

few weIl understood processes which are in prillCIple fully descfl!Jpd by qllrlntllm C'/l>c

trodynamics. In comparison, hadronic showers are mllch morc cornplicatcd, illvolving

strongly interacting particles undergoing nuclcar proce'lscs. A great d<'al of ('ffort was

devoted during the mid to late eightics to undcl stand the ullderlyillg plO('(':-.ses 0«'\11'

ring in hadronic showers. Through detailed Monte Carlo ~tudie,>[7][8J[91l'>iJllulating

the development of hadronic showers, sorne undcrstanding of the contributions ln the

measured signal from the various shower components has bccn achicved.

14

(

(

CIlAPTEH 2. THE CALORl.\fETER 15

2.1.1 Sampling Calorimeters

Sarnplillg calorimetcr5 are composed of inlerleaved layers of passive absorber layers

illld i1ctIVf' d('\(·cto!" lctyf'rs Becau5c materials suited for deteeting particles telld to

he lllad(· hom Ilgh!. eh'llH'nts, alone they would require a very large volume to stop

partlcle5. The presence Jf absorber layers made from high Z clements remedies th!.,

by eau,>lIlg a ~hower to dcvf'lop more quickly. This allows sampling calorimcters to

1)(' COlllpact, saving much in cost and space compared to homogcncous calorimeters.

Also, by corr('ctly sclecting the matcrials and the thicknesses of the absorber and

art ive layers, il is posc;ible to obtain an equal response from the electromagnetie and

hadronic compûncnts of a shower. This tums out to be a criticaJ requirement for

achieving good hadronic energy resolution.

Sampling calorimeters, however, introduee fluctuations III the energy measure

lllf'nt, dt'grading th('lr resolution. These intrinsic sampling fluctuations oecur because

thl' 5hower is sampled only in the active layers; 1 ('. only the energy deposited in

the <Hti"c layers plOduces a measurable signal. The sampling fluctuations arc mostly

the fluctuations in the number of charged particles crossing the active layers. The

sall1pling resolution, O'.amp, is proportional to the square root of the thickness of

th(' sampling laycrs. In eledromagnetic (E~I) showers sampling fluctuations are the

largest colltributors to the energy rcsolution.

Also contributing ta sampling fluet lat ions arc Landau and path length fluctua

tions. Landau fluctuations O(cur because a particle can deposit, through the produc

tion of energet ie delta clectrons, mueh more energy in an active layer that the average

dE/dx 10ss through ionization. Path length fluctuations are due to the multiple scat

terings of a charged particle from Coulomb interactions with the nuclear electric field

which ean cause it to travel a long distance in an active layer and therefore deposit

more energy in there.

üther possible contribu!'ors to the resolution are energy leakage out of the calorime

ter, dctcctor imperfections, noise, photoelectron statistics and pileup effeets.

1 r " [

1 r. >, "

CHAPTER 2. THE CALORIMETER IG

2.1.2 Electromagnetic Showers

When a photon, dectron or positron cnters a calorimeter it will en'ate an dedrol11ag

netic shower. The main processes involved are the following:

• Bremsstrahlung: In the electric ih'ld of an atom a high energy electron call be

deflpcted and emit an energetic photon. The cllergy loss pel' unit length g()es

as Z2.

• Pair production: A photon with at lcast twice the l'est CIlC'l'gy of an (·kctl'oll

(1.022 MeV) transforms into an electron-positron pair. This cali only (lCt·tIr in

the presence of a third body, lIsually a nucleus, to conserv(' mOIl1Cllt UHl. 'l'hl>

cross section gocs as Z2.

• Compton Scattering : The photon loses encrgy by the in<,lastic scat t('ring off

atomic electrüns. The cross section is pwportional to Z.

• Photoeledric effect : The photon is totally absorbcd by an atornie c\cc1 ron,

freeing it from the atom. The cross section increa.<;cs as Z4 or Z5.

• Ionization : Chargcd partides ionize elcdrons from neighbouring atollls, pro

ducing secondary eledrons and exeÎted atoms. The pnprgy los ... pel' unit lt'lIgt.h

go cs as Z log Z.

From Figures 2.1 and 2.2 we can see that an EM sl1o\\'CI' will go through twü

stages. In the beginning energetic pal'ticles 108e cIlelgy thro\lgh br('msstrahhmg aud

pair production. This continues until the pal tide rcadws a (,PI Utin thrp"hold CIWlgy,

called the critical energy €c, at which the energy loss to brems~j rahlllTlg ih ('qu,t! 1.0

the ionization Joss. Afterwards, in the later stage of the S}lo\\W, "ItOW!'1 parti!]P~ Jo,,('

their energy through ionization bdorc bcing able tü producc ncw pal ticlcf, dnl! hhower

mu1tiplicatioll stops. This ionization energy dcpo!:>ited by lo\\' cnergy paIl icl('~ al, 1 be

end of the shower constitutes a large fraction of the measUIcd signal. Th(· CI il.Î( al

(

(

CHAPTER 2. THE CALORIMETER

-• -. , .. ..J

"'1-'0'0

-114.1 ,

E (MeV)

020

Ol~

{ u

010 ~ c ::1

OO~

1000

17

Figure 2.1: Fractional Energy Loss for Electrons and Positrons in Lead (from Rev. of

Part. Prop., Phys Let. B. Vol 299)

••• '.H

• •.• l '.14

'.'2

Figure 2.2: Photon cross-section in lead as a function of photon energy. (from Rev.

of Part. Prop. 1980 edition)

-

CHAPTER 2. THE CALORIMETER 18

energy le can be parametrized by [10]

f c :::= 800 [ AI e V]. Z + 1.2

We describe the longitudinal and transverse devclopmcnt of a showpr in terms of

radiat,ion lenGth and of Molière radius. The radiation length, Xo, of a mat<'rial is tlH.'

distance a high cnergy electron must travel for its energy to drop to 1 je of its original

energy purely due to bremsstrahlung. It can be approximated hy [11]

The average distance a very high energy photon trave1s before splitting into an e+ e

pair is 9/7 Xo. The Molière radius, R.u, is a measure of the radial devdopnwnt of

an EM shower. Most of the contribution to the radial spread cornes From multiple

scattering of the electrons. 95% of the total energy of a shower is contained in a

cylinder of radius approximately 2 RAI. It is roughly givcn by

For a DU-scintillator calorimeter, the Molière radius is about 2 cm.

The average amount of energy deposited at a given depth from the face of a

calorimeter by EM showers can be parameterized according to [121

dE ba+1

E a -bz dz = r(a + 1) z e .

Here the constants a and b have a logarithmic dependence on energy :

a = ao + al ln E

b = bo + b1 ln E,

where for the ZEUS calorimeter the values are: ao = 1.1,5, al = 0.54, ho = 0.395/ X o,

bl = 0.022/ X01 E is in GeV and Xo is in cm [13]. Longitudinal shower profiles are

shown in Figure 2.3 for electrons of energy l, 20 and 75 CeV.

{

CHAPTER 2. THE CALORIMETER

J r 0" .. t 0'.

012

0'

001

001

oa.

002

°0

1

1

. .. . ..

1

1

1

.. 1

'. ............ ........ \1 20

19

21

Figure 2.3: Average deposited energy vs depth for electromagnetic showers of energies

l, ~O and 75 GeV

2.1.3 Sampling fraction

The sampling fraction, defined as the fraction of a particle's energy deposited in the

active layers, plays an important role in determining a calorimeter's resolution. This

sampling fraction, R, of a particle is given by

R - Evu - , E"j, + E,Av ••

where Ev;. is the energy deposited in the active layersj Etrun.. is the ellergy deposited in

the passive layers. The different types of particles produced in a sbower have different

sampling fractions. For electrons and hadrons we denote the sampling fraction by e

and h respectively. Sampling fractions are uS'lally compared to those of minimum

ionizing partides (mip). These ficticious particles have a sampling fraction given by

wbere i runs over ail tbe different media; tl is the thickness of a layer of medium Îj

dEddz is the minimum energy loss per unit length in medium i, and act is the active

medium. Muons have a sampling fraction close to a mip and are often used as mips.


For an electron the cjnllp ratio is less than one. The rt'é\son is due ta the dif

ferent Z dcpendence of the cross sections of the various intclactiol1s oCl'urril1g in an

electromagnetic sho\\'er. ln the course of an e}ectl'omagnctÎc show('r m,Hl!, low t'Ill'rgy

gammas are produced. For these photons the photo<'lcctri<: df<,ct, is the dominant.

interaction. Since the cross section of the photoclectric e{fcct gacs a.s Z5, t IH'y are

predominately absorbed in the high Z absorber rcgion. The resulting 10w elll'rgy

electrons have a very short range and mO!,t do not rearh the scintillator. The e/mip

ratio for a DU jscintillator calorimeter is about .62 and is ollly slightly depende/lt on

energy.

2.1.4 Hadronic Showers

Hadronic showers are created through the inelastic scattcring of the strongly int.eract

ing hadrons and their sccondary particles by the nuclei of absorber llHÜ<'l'ial. A wide

variety of different particles are produced, with differing energy 10ss '_H'chanislns. A

summary of the particles found in the cascade is provided below.

• Charged hadrons such as ]('s, 7I"'S, and p's which lose energy through ionization.

• 1/"°'S and ",'s produced from the decay of hadronic resonances. They d('cay into

two ')"s and deposit their energy in the fonn of electromagnetic showers.

• High energy neutrons produced in the intranuclear cascade during spallation.

• Low energy neutrons released through evaporation of excitcd nuclci.

• Low energy gammas produced in fission processes and thermal neutron capture.

• Neutrinos from particle decay.

We can define a nuclear intera,·tion length, À in a similar manner to to the radiation

length. >. is the average distance a hadron travels before colliding with a nucleus:

(

c


where NA is Avogadro's numbcr; A is the atomic weight of the nucleus, and O'abJ' the

nuclcar ab~orption cross section, is proportion al to A711 [14).

Then the average longitudinal profile of the energy deposition in hadronic showers

can be wriUp!l as

where a and b have the same values as in EM showers[15J. For a model of the ZEUS

calorimeter a = 0.16, 9 = g1 + g21n E, g1 = 2.86/ À and g2 = -0.50/ À and À = 23.6

cm.

The transverse dimension of a hadronic shower grows logarithrnically with the

cnergy of t.he shower. The width, W, needed to contain 99% of a shower in an iron

Iiquid argon calorimctcr Îs [16]

W(E) = -17.3 + 14.31n E [cm].

The particle diversity causes the rcsolution to be worse thall pure electromagnetic

showers. Each type of particle has its own sampling fraction. Neutrinos and sorne

of the neutrons will totally escape the calorimeter without being detected. AIso,

the energy loss in breaking up the nuclei is not seen. Heavy charged particles will

dcposit cnergy through ionization si!TIilar to a mip. The lI'°'S and the 1]'S have only a

fraction of a mip's signal. The fluctuations of the proportions of particle types and

the intrinsic fluctuations of hadronic showers are the main contributors to the energy

resolution ~f hadronic calorimeters. The extra undetected energy losses cause el h to

be usually greater than 1.

A large source of fluctuations in a hadronic shower is the fraction lem which is

in the form of an electromagnetk shower. This fraction can vary widely from event

to event with the production of high energy 1I'°'S and 1]'S early in the cascade. If the

response from the electromagnetic and hadronic components of a shower are not the

same, the resolution of the calorimeter to hadronic showers will be seriously degraded.

Moreover, f~m is energy dependent and non-Gaussian.

..


2.1.5 Compensation

Due to the presence or the electromagneti-.:ally decaying 1r°'S and 1] 's, hadronic showcrs

have both an EM component and a nonelectromagnetic component. If tlH' elh r:itio

is not l, the non-Gaussian fluctuations in Itm will cause CTZ,J "7 ta Ilot impro\'c:' as

liVE. At high energies the resolution will approach a llonzero constant tt'I'm. ln

addition, the tnergy dependence of lem causes el h to be 3180 cnergy dt'jwndent ThIs

lesults in an alinearity in the calorimcter signal which cO\llcl bias tI iggers La.::;ed 011

enelgy because the detected energy of a single encrgeti(' partide wOllld he cliff('n'Ilt

than that of a jet with the same total energy. 19noring dctcdor imp(" [ections, tht'

resolution of a hadronic calorimeter can gcnerally be writt('ll as

~ = A/tabll + B(ejh - 1) E JE '

where tabs is the absorber layer thickness, A = (T2 mtr + (T2 Mmp) t is tllP contri bu

tian from intrinsic fluctuations, nuclear binding losses and sampling fluet uat.ions and

B(O) == o. In order to achieve ej h == 1 we have to compensate for undetected energy losses

due prirnarily to the breaking up nuclei. We can either suppress the calorirnetcl"~

response to the electromagnetic component of the shower or incrcasc the respOllS<'

to the hadronic component. Most of the compensation is usua l1y attélJned through

augmenting the hadronic response. For example the use of deplcted uranium platc!>

increases the hadronic response due to the extra en~rgy released during fission pro

cesses. However, the most significant factor in increasing the hadronic rcsponse is

the a.mount of energy neutrons transfer to a hydrogenous medium. It hll'ns out that

in the final stages of a shower's development most of the showcr's energy is sp('nt on

nuclear processes. Low energy neutrons, protons and gammas arc produccd. Many

more neutrons than protons are released, especially in large Z matcrials. Thes/' soft

neutrons are eventually recaptured and lose their energy invisibly through e1a:,tic

scattering. However, if the calorimeter contains hydrogen in the readout matcrial,

much of this lost energy will be l'ecovered. The neutrons can transfer a large propor-

(

Cl/APTER 2. THE CALORIMETER 23

tion of their kinetic cnergy to hydrogen, prorlucing recoil protons. These low encrgy

(l'V 1 MeV) protons have a very short range and becausc they originate in the active

meltcrial, they are Ilot sam pIed. Instead, they will deposit ail their energy in the ac

tive calorimeter layers. However, saturation of the medium can limit the extra light

contribution from the ionizing protons. This is parametnzed by Birks law,

dL == A dEldx [cm2 j g). dx 1 + kBdEjdx

where L is the light yieldj A is the absolute scintillation efficiencYj kB Îs a parameter

relating the dCl1'3ity of ionization centers to dEI dx. Ncvertheless, the net result is a

decrease in the el h ratio. The amount of compensation can be chosen by varying the

relative thicknesses of the active and passive layers of the calorimeter, or by adjusting

the fraction of hydrogcn atoms in the sampling medium. A final adjustment can

be done by changing the signal integration time which determines what fraction of

delaycd gammas from neutron capture contribute to the signal. The elh ratio can

also be reduced by lowering e. This is done by inserting a low Z foil between the

active and passive laycrs. This prevents photoelectrons produced at the boundary of

the high Z material from reaching the a,ctive layers.

Compensation has been achieved for uraniumfscintillatoI and lead/scintillator

sampling cc:tlorimeters. In order to be compensating, a DU/scintiHator calorime

ter must have a scintillator/uranium volume ratio of about 0.82. The volume ratio

is about 0.25 for a scintillator/lead calorimeter. This means that a compensating

leadjscintillator calorimeter of the traditionallayer type sampling variety would Juf

fer from larger sampling errors than the DU Iscintillator \'ariety if normal scintillator

thicknesses were used. Partial compensation has been achieved using liquid argon.

In this case neutrons transf~r their energy tü argon in the same way they do with

hydrogen, but with mu ch less efficiency. At most only 10% of the neutron's energy is

given to the recoil argon atom and this also occurs with a lOWf'f rrn'lS section[8]. It is

also possible to improve the rcsolution of noncompensating calollllld ('rs by estimat

ing the fraction of energy deposited through 11"0 deray and then appIj illg weighting

algorlthms. This would require a fine longitudinal separation in the readout and was

CIIAPTER 2. THE CALORIMETER 24

fil'st attempted by the CDHS expel'iment[lï].

The hadronic t'nergy resolution of a compensating DU j~ciJlt illatol' calorillwtel' is

thc quadratic sum of its illtrmsic and sampling l'csolution :

a(E) 22% 0.09J~E(1 + l/Npe ) -- = --ffi --E JE vIE '

where E is in GeV, ~E is the energy loss pel' layer in l\1eV and Npe is the numbel' of

photo-electrons seen in the phototubc. This yiclds an energ}' r('solution fol' a typical

calorimeter of a(E)jE = (33% - 35%)jJE.

2.2 The ZEUS Calorimeter

Motivated by the need for a high resolution calorimder to satisfy the physirs 1'('

quirements, the ZEUS collaboration began u.'dertaking in 1985 d('sign studies for

a depleted uranium - scintillator calorimeter. Only after an extensive plOg,ratn with

test calorimeters and Monte Carlo studies was the design fin \lizcd [18). The sa III pli ng

thickness in the EMC and HAC section was chosen to be 1 .xl. leading ta a DU plate

thickness of 3.3 mm.

The DU plates are fully encapsulated by a stainlcss steel foil of 0.2 mm thickn ... ss

in the EMC and 0.4 mm in the HAC. The steel foil a11owe(1 saCe handling of tlw DU

plates during construction (the main concern was uranium dust), as well as a reduction

of the signal contributed by the DU natural radioactivity. This radioactive signal

(UNO) has to be low to minimize noise and radiation damage to the scintillators but

large enough to be used for intertower calibration. The choicc in DU plate thickm'ss

required a scintiUator layer thickness of 2.6 mm to achieve efh = 1.

The scintillator used is SCSN-38. It has a high light yield, low light attenuation

and good stability against aging and radiation. The use of plastic scintillator with its

fast decay time of the order of a few nanoseconds allows a very good timing rcsolution.

This is important in reducing the background from cosmic ray and bcam gas evcnts.

The fast response also alleviates pileup problems arising from the short interbullch

(

CJJAPTER 2. TJ1E CALORIMETER 25

crossing time. The composition and properties of the sampling layers are presented

in Table 2.1.

The EMC sectior. is made of 25 DU /scintillator layers. It is followed by the HACI

section and ,in FCAL and BCAL only, the HAC2 section, each with up to 80 layers.

Aside from the 60Go sources, the calorimeter uses two other calibration systems. The

UNO mC/ltioned above allows the normalization of the signal from towt'r to tower

without having to subject every tower to a test beam. Test results have shown that

e/UNO (the ratio betwcen an electron's signal at a given energy and the UNO signal)

betwf'cn diffcrcnt calorimcter towers is constant to wit hin 1.1 % in FCAL and 1.5%

in RCAL[19]. The other calibration system uses pulsed laser light fed to the base of

the light guide by optical fibres to monitor the linearity and long term stability of the

photomultiplier tubes.

The calorimeter subcomponents are divided into modules. BCAL is made from

32 identical modules. FCAL and ReAL each contain 24 modules, two of which are

half and positioned above and below the beam pipe. A summary of the dimensions

of the modules in FCAL and ReAL is given in Tables 2.2 and 2.3.

Each FCAL/ReAL module il; 20 cm wide and has a height varying from 2.2 to

4.6 m depending on it:; pOE: ".l11 to the beam. Thus the number of towers in a module

ranges from Il to 23. Figures 2.4 and 2.5 shows the positioning of the FeAL and

ReAL modules.

2.3 FCAL/RCAL Modules

The modules for FeAL and ReAL were assembled by the Netherlands and Canada

[19][20][21]. Each consists mainly of a stack of depleted uranium plates and scintillator

tiles supported by a steel C-frame (see Figure 2.6). The C-frame has three parts, the

end beam and the upper and lower C-arms. The end beam supports the stack of

scintillator and DU plates during assembly and transport. Running along the end

beam are two trays housing the optical fibre bundles for the laser system and the


material thickness thickness thickness [mmJ [.\oJ [ÀJ

EMC steel 0.2 0.011 0.00i2 DU 3.3 1.000 0.0305 sted 0.2 0.011 0.0012 paper 0.2 scintillator 2.6 0.006 0.0033 paper 0.2 contingency 0.9 sum 7.6 1.028 0.0302 effective .\"0 0.74 cm effective À 21.0 cm effective RAI 2.02 cm effective (e 10.6 ~leV effective p 8.7 gjcm3

HAC steel 0.4 0.023 0.0024 DU 3.3 1.000 0.0305 steel 0.4 0.023 0.0024 paper 0.2 scintillator 2.6 0.006 0.0033 paper 0.2 contingency 0.9 sum 1 8.0 1.052 0.0386 effective Xo 0.76 cm effective À 20.7 cm effective RM 2.00 cm effective (e 12.3 MeV effective p 8.7 gjcm3

Table 2.1: Composition of a Sarnpling Layer in EMC and lIAC

l


FCAL module type FIT FIB FJ1 F12 F2l F22 F3 F4 F5 F6 no. of modules 1 1 2 8 2 2 2 2 2 2 ~;ve helght (cm) 220 220 460 460 420 420 380 340 300 220

no. of 20x20 cm2 t.owers 11 11 23 23 21 21 19 17 15 11 no. of 5x20 cm2 EMC sections 36 36 76 68 52 44 36 12 no. of 20x20 cm2 HACO sections 2 2 4 6 8 10 10 14 15 11 no. of 20x20 cm2 HAC1,2 sections 22 22 46 46 42 42 38 34 30 22 no. of EMC channels 72 72 152 136 104 88 72 24 no. of HAC channels 48 48 100 104 100 104 96 96 90 66 maximum depth (À) 7 1 7.1 7 1 7 1 7 1 71 64 6.4 5.6 56

Table 2.2: Summary of the dnnensions of FCAL modules

RCAL module type RIT RIB RU R12 R21 R22 R23 R3 R4 R5 R6 no of modules 1 1 2 6 2 2 2 2 2 2 2 active height (cm) 220 220 460 460 420 420 420 380 340 260 220 no. of 20x20 cm2 towers 11 11 23 23 21 21 21 19 17 13 11 no of 10x20 cm:! EMC .;ections 9 18 38 34 30 26 22 18 6 no. of 20x20 cm2 UACO sections 2 4 6 6 8 10 10 14 13 11 no. of 20x20 cm2 BAC1,2 sections 11 11 23 23 21 21 21 19 17 13 11 no. of I!:MC channels 18 36 76 68 60 52 44 36 12 no. of HAC channels 22 22 54 58 54 58 62 62 62 26 22 ma:dmum depth (.\) 4.0 4.0 4.0 40 4.0 4.0 4.0 40 4.0 3.3 3.3

Table 2.3: Summary of the dimension!> of RCAL modules

c.

r 'i.

r , , ,


J s.s~ 1--~ .... ~.

10- r- fo-~ f0-

r-

l-i- -

1 71 ).

6 5 4 3 2 2 1 1 1 1 ~ ,,!. 1 Il l, 2 2 3 (, S 6

lB

L-~

'-t-

l.-

I Il Il !

110 ., QI .. 1 1""

Figure 2.4: Positioning of FeAL modules

~~ -~- ~- 33>-

~ ~ r-- 0-

r-1-

~ r-

6 5 4 3 2 2 \ 1 , 1 ~ If.! , 1 1 1 \ 2 2 3 (, 5 6

18

,.0)'

.... 1-

L-....

'-

'10 ., III • • CM

Figure 2.5: Positioning of ReAL modules

CHAPTER 2 THE CALORl.\lETER 29

Lrass pipes for GOC'o calibration. Access to the fibre bundle and source pipes is via

the upper C-arm. When the module is finally installed it rests completely on the

lowcr C-arm.

2.3.1 DU /scintillator stack

Depletcù uranium is an alloy consisting of 98.4% 238U, $ 0.2% 235U and 1.4%Nb. It

lias a d<!lIsity of about 1~.9 gjcm3 with a radiation length of 3.2 mm and a nuclear

absorption length of 10.5 cm. The DU plates are 3.3 ± 0.2 mm thick, and 188.8 mm

wide in the ENtC section, 183.8 mm in the HACI section and 178.8 mm in the HAC2

section. The plates were limited by production methods to only '" 3 m long so it

was Ilcccssary ta weld two together for the larger modules. They are surrounded by

a layer of .2 mm thick steel foil. Those in the HAC sections have a second layer ta

furthel reduce the DU noise.

At the front of the DU /scintillator stack is a 15 mm thick aluminum front plate

with the edges smoothed to allow the steel straps compressing the stack to slide.

Immediately behind the plate lies first a scÎntillator layer and then 25 layers of

DU jscintillator of the EMC section. After the fourth scintillator layer (and also

aft(\l' the 7th in FCA L) is a 15 mm gap for the silicon HES detectors. A thin steel

shect kceps the flcintillator tiles from falling into the gap. The first scintillator tile

has no DU plate bdore it bccausc the dead rnaterial in front of calorirndcr (from the

solenoid, tracking detectors, and front plate) already contributes about a radiation

lengt h of absor ber.

The scintillator material, SCSN-38, is an aromatie plastic with a cross linked

polystyrene base doped with two wavelength shifting dyes, butyl-PBD (1 %) and BDB

(0.02%). The scintillators tiles were saw cut and machine polished on an four edges.

They have a thickness of 2.6 mm ± .2 mm. This thickness distribution is shown in

Figure 2.7. Thr tiles \Vere selected sueh that the those closest to 2.6 mm thick were

used in the EMC sections. This was to help give the EMC sections, which had only

26 scintillator layers, the best uniformity. The 4 EMC tiles in a layer of a FeAL


~

~ .. ~

J ~ " .. il j

! :;:

~

Figure 2.6: View of a Large FeAL Module

--

f


... .... III FEY': 1ft •• "" ...

.. FHACI ~ rm ... 1 l " .!: 'MO

.-=- ç. .OIO

IlOG

j MO

• \ • ,. •• ., ., Il ••• 1 .• 1.' 1 .. '.t ,",Ic:kn" .. (mm. Thlch ... (lIua)

Pl' 10. '1' IOc

•• ... '" FHACO ............. .. ~- " rml-Ill' ... ...

{j .-... - .. ••

1 ' ... 1.1 U 1. 1 .. .. ••• •• U • •• ,",Iellne .. (111111. Th'ch_ ( •• )

Fl, lOb ri, lOcI

Figure 2.7: Thickness Distribution of FeAL scintillators

tower are labelled EMCl, EMC2, EMC3, and EMC4 with EMCl being closest to the

bottom C·arrn. In RCAL the EMC tiles are similarly labelled EMCl and EMC2.

The DU plates are kept separated by tungsten carbide spacers positioned every 20

cm along the edge of the plates. Because there is less weight to support in the EMC

sections, it was possible to use there titanium carbide spacers ",hose lower Z reduces

their effect on shower development. To accommodate the spacers aU the scintillator

tiles except the EMC2 and EMC3 tiles in FCAL have cutouts at the corners.

To improve the uniformity of response from the scintillator tites, they were wrapped

with white and black paUerned Tyvek paper 0.18 film thick. The Tyvek paper is kept

in place by two black Tedlar sleeves at the readout edges. This keeps the uniformity

of response to within ± 2% in the EMC and ± 4% in the HAC over 95 % of the

scintillator area. Figure 2.8 shows the pattern used for the FCAL EMC scintillator

tiles. The patterns were coarse enough such that only one type was needed for each

type of scintillator.

..... . ,' ...


Figure 2.8: Tyt·ek Paper Pattern used for FCAL EMC Scintallators

2.3.2 Optical Readout

The optical readout consists of WLS plates and light guides transporting the scintilla

tor light to the PM tubes where it is converted to an electrical signal (see Figure 2.9).

The WLS is a 2.0 mm ± 0.2 mm thick plate made from polymethyl methacrylate

(PMMA) doped with the fluorescent dye Y-7 and an ultraviolet absorbant. The UV

absorbant culs off wavelengths below 360 nm. This was intended ta reduce the con

tribution to the signal from Cerenkov light produced by showers occurring in the

cracks between modules. In the EMC and HACO sections the Y-7 concentration is

45 ppm. AU other sections use a concentration of 30 ppm. The higher concentration

of Y-7 in the EMC WLS increases the light yield by absorbing more of the scin

tillator light belore il passes out of the WLS and scatters off the back refledor (see

below) The Y -7 dye allows the WLS to bring sorne of the scintillator light through the

ninety degree turn to the PM by absorbing the blue scintillator light and re-emitting

it isotropically[22]. The light guide at the end of the WLS is made of bent strips

joining the rectangular end of the WLS '0 the circular surface of the PM photocath

ode. The light guide transports the light from the WLS to the PM adiabatically by

transforming the cross sectional surface of the WLS from a long thin rectangle to a

(


fi l S r"AC 1 fi l S fEwc/rHAC 0

l <. 'II" ' ,. l .. f .. t.( J

Figure 2.9: View of WLS reading out scintillator

square matching belter the photocathode surface, while maintaining the same cross

section al area. The light guide and WLS plate are made as one piece in order to avoid

light losses from glue joints.

To help optimize the uniformity of response 'end reflectors' and 'back reflectors'

made from aluminized foil were add,~d. The end reflector positioned at the end oppo

site the light guides reduces the attenuation and significantly increases the totallight

yield. The response of each wavelength shifter was measured and an individualized

pattern of either black dots or black lines was applied to the back reflector to reduce

the position dependence. The final uniformity of the WLS was within ±3%. The

WLS plates reading out aIl the sections for one side of a tower are mounted in a cas

I)dte made from 0.2 mm thick stainless steel. Between the EMC plates lie the brass

tubes used in the cobalt scanning, 2 per cas8ette in FeAL and only 1 per cassette

of ReAL. Foam rubber between the c:assette and WLS pushes the WLS against the

scintillator tiles. However, direct contact between WLS and scintillator which could

cause partial optical contact is prevented by 0.4 mm thick nylon fishing lines.

In the final stage of the optical readout system the WLS light is collected by the

photomultiplier tube. A PM consists of a photo-cathode which emits electrons when

1

-


st.l'uck by photons. These 'photoelectrons' are accelerated by an electric field through

a series of dynodes set at increasing voltage. Upon striking a dynode an e}Pctron

will rclease many more electrons which are aIl 3cc("lerated ta the next. oynoof>. This

multiplicative effect amplifies the signal such that only a few photons ar(' 11('('d(·d tü

strike the photocathode to produce a measurablc signal. The PM t.ubes are l'equia'd

to have good gain stability, good linearity over the entire dynamical range and small

dark current. The XP1911 tube from Philips was chosen for the EMC sect ions of

FeAL and the R580 tubes from Hamamatsu Photonics for the l'est. Both types a.re

10 stage head-on tubes with diameters of 19 and 38 mm respectively. They have bi

alkaline photocathodes with a spectral rcsponse matching well the spect r\\m of Y-7

dye in the WLS. The tubes are equipped with Cockcroft-Walton bases to supply the

high voltage. These bases step up the voltage internally, dissipating less heat than

resistive bases and requiring a maximum external voltage of only 24 V.

2.3.3 Straps

The DU jscintillator stack and WLS cassettes are kept together by .25 mm thick

steel straps each 196 mm ".ide. one for each towcr. The straps are tcnsioned to a

force of 15-20 kN per tower. The straps press onto the aluminum front plate and

are fixed below the frontplate of the endbcam via bulkhcads to the photomult,iplicr

housing assembly. The gaps between the straps as weIl as cracks Icading to the optical

readout system were covered with metal tape and black tape to prevent outside light

from leaking in.

f

(

Chapter 3

60Co Source Scanning

The short term goal of the 6OCO scans was to discover construction imperfections

which would affect the response uniformity of the calorimeter. This was possible by

ta!~jng advantage of the relatively limited range of the emitted gammas which illu

minated only a few layers at a time. Local information about the structure of the

calorimeter stack and the response of the WLS could be obtained. The scans were

done beforc installing the modules into the ZEUS detector in order to have time to re

paie them, if possible. It will also he used as a long term monitor of the eharacteristics

of the optical components such as the WLS uniformity and scintillator attenuation

length. Plastics, used as the base material for scintillator and WLS, are vulnerable to

damage from aging and radiation, causing them to turn yellow and producing minute

surface cracks (crazing) which degrades their optical qualities. This decreases their

attenuation length and reduces their light yield. Because the scintillator layers clos

est to the interaction region receive more radiation, a variation in the light output

efficielley from one layer to the next can oceur. Snch a nonuniformity in the optical

l'eadout, whether from radiation damage or intrinsic ta the construction, would in

erease the sampling fluctuations and worsen the energy resolution. More importantly,

sinee the longitudinal shower profiles are energy dependent, the nonuniformities could

cause the absolute energy calibration of the calorimeter to be nonlinear. For example,

i t h as heen est imated that to keep the barrel calorimeter response variation un der 1%

35

CHAPTER 3. 6OGO SOURCE SCANNING 36

for eledrons in a range of energies 1-100 CeV in the EMC section, th(' nOlllllliformity

in the WLS must be less than 7.2%.[23J

In or der to condllct scans with radioactive SOUfces, the sour('(' 1l111",t somc!\Ow h<.>

brought into close proximity with the scintillators. Since the modules. Ollce installed.

are packed as closely as possible side by side to minimize dcad SpilCC, the dcli\'<'l'y

system has to be interior to the modules if any scans arc to made after the ZEUS

experirnent begins. Because there is no space available within t.he DU(scintillatol'

stack (drilling holes in the plates and tiles would have had serious ::-;ide ('ff('d.s), the

source has to travel eithcr in or be~ide the \VLS casbettcs. Tu!ws are ail ohviollS

choice to use to guide the source. If we want to scan each EMC sect ion III tlte saillI.'

wa)', a lT'inimum of two tubes (each between a pair of EMC WLS) are f('<JuÎI'<'d »('1'

FCAL cassette and on1y one per RCAL cassette. The tubes wllich rlln clown tlH'

WLS must somehow be accessible from the outsidc. This is only possible through

the upper C-arrn. Therefore the tubes are forced to undergo a ninety degr<'(' tUl n in

order to reach the C-arm. Using the C-arm does however offer the advantagc that

ail tubes can be accessed from one location in the mudule. The source itsc\f Rhollid

be small to provide fine detai! on the layer structure. To bring it in and out of the

tubes, it has to be attached to a wire or rope of sorne sort. The 90° bcnd in the tubes

requires a wire flexible enough to bend without kinking, but stiff enough to Pllshed

through a vertical bend. These were the design considerations.

In the end two diffcrent procedures were used in conducting source scans, one

with the source inside the module (insidc scanning), the other with the source outsidc

(out si de scanning). At first , inside scanning was performed at CERN and at. DESY

on a few large FeAL modules, where a radioactive 6(JGo source at the end of a piano

wire was inserted into tubes running along the WLS inside the module. The signais

from the PMs were read out as the source wire was slowly pulled out by a computer

controlled driver. After an accident with the inside scanning, adjustnH>llts \Vere made

80 that aIl modules were scanned on a preparation stand by a source wire inserted

into tubes rnounted on a movable platform outside the module.

(

c

CHAPTER 3. 6OCO SOURCE SCANNIl\'G 37

60 CO was chosen as the radioactive source to scan the modules because of its

energctic gamrnas, long lifetime and ease of handling. It emits gamma rays at two

energies, 1.173 and 1.332 MeV, in equal proportions. Gamma sources are superior to

clectron sources because low encrgy electrons have a very short range and would stop

in the WLS bcfore rcaching the scint.illator. A lower energy gamma source would have

the advantage of illuminating fewer layen. at at timc, but the reduced penetration

depth could also increase the scan 's sensit.ivity to effects arising from scintillation

light being produced very close to the scintillator edge. 6OCO'S long lifetime of 5.271

years rclievcs the necessity of frequently making new sources. Only one source was

needed to scan aIl the modules over a period of six months.

3.1 Physics of Light Production

In the energy rcgion around 1 MeV, the dominant interaction of photons is Compton

scattering (cJ. Figure 2.2). The energy transferred, Ek, to the electron in Compton

scattering can be expressed as a function of the scattering angle, B, by

For a "t with an energy of 1.332 MeV, the maximum energy transferred is 1.038 MeV

(for 0 = 180°).

Sorne of the energy deposited in the WLS by the fast electrons produced from

Compton scattering is re-emitted by the excited atoms in the form of Cerenkov light.

Electrons travelling in a dielectric material wit.h an index of refraction n will produce

Cerenkov light if their velocity is greatcr than c/n, the specd of light in the medium.

The amount of energy emitted as Cerenkov light with a wavelength, -X, is given by

where re is the classical radius of the electron. Most of this light is absorbed by the

ultraviolet absorber in the WLS, but sorne of the longer wavelength components may

CHAPTER 3. 6OCO SOURCE SCANNING 38

eventually iravel down the WLS and contribute to the signal in addition to that {rom

the scintillator light.

AIl organic scintillators contain a benzene ring in thcir structure. The iOllizing

energy from a passing high energy particle excit.es free valence electrons. Th<'y t hen

lose vibrational energy and the molccule enels up in a vibrational l('vel of the fhst.

excited state without the emission of light. After a short period of lime the mo!ecu!t·

ernits a photon, decaying to a vibrationallevel of the grùund statc b(>fore falling to the

ground state. The net resuIt is that scÎntillator is transparent tü its own rad.iation.

The two dyes used in SCSN-38 inCl'ease the attenuatiÜll !ength as weil dS shift th<,

ultraviolet light from the base to a wavelength matching the absorption spectrull1 of

the Y-7 in the WLS, Figure 3.1 shows the absorption and cmission spcctfi\ of the

dyes involved,

The light then travels via total internaI reflection to one of two ends where it

enters the wavelength shifter. Only light irnpacting the boundary bctwcen air and

scintillator at an angle greater than 0 = arcsin(nt!n2), where nI and n2 are the indiœs

of refraction of air and scintillator respectively, will be reflected. For SCSN, n = 1.48

and () = 42 degrecs. Of the light that i8 not reflected and passes out of the scintillator,

most will be partially scattered by the Tyvek wral_ping. Sorne of it which r('('ntNs

the scintillator rnay be absorbed by the dyes and re-emittcd in a dir(,ctiotl sllch that

it eventually l'eaches the WLS. The patterncd back reflector of the WLS works in a

sirnilar way as the Tyvek paper. Sorne of the blue !ight frorn the scintillator pa.,>scs

through the WLS without being absorbed and re-emitted isotropicall~ tlS green. The

pattern on the back reHector controls how rnuch of this Iight is rcflccteJ back into the

WLS to have a second chance to be absorbcd and rc-emitted.

3.2 Inside Scanning

In FeAL the EMC towers have 2 brass guide tubes pcr side, one rUIlning between

the EMCl and EMC2 WLS, the other between the EMC3 and EMC4 WLS. For

1

CHAPTER 3. 60CO SOURCE SCANNING 39

the HACa towers the tubes run along the spacer column, and for REMC they run

bctwccn the EMC WLS strips. The tubes have an outer diameter of 2.5 mm and an

inner of 2.0 mm. The tubes run straight to near the end of the light guide where

they are each glued to an elbow. Figure 3.2 shows the region around the light guides

where the elbow is located. The elbows join to long brass tubes lying on a tray along

the web plate of the end beam. These tubes run towards the upper C-arm where

they grouped together in an interÎace. A fanout lies in the top C-arm connecting the

interface via brass tubes to holes arranged in a circle of a cover plate to which the

driver is mountcd. The driver is first attached to an extender which allows the driver

to he located on the ReAL modules which are rnost restricted in space. The extender

is an aluminum plate with a ch'de of short brass tubes projecting out. These tubes

link with holes in the base plate of the driver. The exlender is mounted to the fanout

by four brass pegs and two long screws. The driver pushes the wire tû the €ùd of a

selected brass tube. It then draws the wire out while data from the PMs is recorded

onto disk.

The driver had difficulty inserting the wire to the end of the brass tubes due to

the buildllP of frictional forces. Often the wire had to be manllally inserted past the

elbow to the end of the brass tubes from where the run wOllld be started. Sorne of

the elbow joints were not properly glued and the source wire would pop them open.

The wire would then exit the elbow and be pushed underneath the straps. This

caused kinking in the wire. The damage to th{' source wire eventually prompted the

development of an alternate intcrim scanning method (see below).

3.3 Outside Scanning

An accident occurred while scanning module NL4 with a repaired source using the

inside tubes. The source wire broke off at the repaired joint and was left behind in the

module when the wire was removed. The source was recovered, but had this happened

alter the module had been installed in the ZEUS detedor, its recovery would have

-

CHAPTER 3. 60CQ SOURCE SCANNING 40

required the removal of the module. Further inside scanning was stlspended until

sorne method to assure that such accidents would not be l'epeated \Vas found.

A new method with using a driver and straight guide tubes 1l101lllted on a movable

platform outsicle the module was implemented in order to continue the sourn' scans.

A modified source driver was attached to a movable platform on roll('rs. 1'h(' driver

was positioned on top (over the aluminum front plate) of a towcr. Two aluminum

wings containing brass guide tubes hung from the platform, one on cach side, clown

the length of a lower. Two different pairs of wings were made, one for FCAL and one

for RCAL. The wings were 20 cm wide and 12 mm thick. The FCAL wiIlgs have -1

brass tubes pel' side spaced 5 cm apart. The ReAL wiogs have only 2 tubes I)('r wing

spaced 14 cm apart. In addition, one wing from each pair has an additional tl1l)(' use<.1

for loading and unloading the source wire. These tub('s had ollly a very slight l)('tl<.1

in going from the driver to the wings and no joints.

For the FCAL modules the platform was positioncd suclt that th<, oUÜ'r brass t.l\h<,s

were positioned o\'er the inner ones. In an attcmpt to position the wings accnratply,

they were fixed in place by alumÎnum profiles screwed to holcs in the PM mounting.

The ReAL wings were positioned by cye using the gaps betwc<,n the straps as a

guide directly over EMC towers or straddling two HACO towcrs. They were fixed in

place by a wedge betwcen the wing and the crossbal's used to ~upport the modL1Je

during transportation. Since the source tubes were 3 cm from the ne'arest end of the

scintillator, a very precise positioning was not necessary.

3.4 The Source Wire

A number of different source designs were tested in the prototype calorimetcr. Two

source wires of the same design were used in the inside measurements. ACter being

damaged one was later modified and used as the source wire for the ol\l~jdc scanning.

The original source wires (Figure 3.4) used in the inside measurements were manufac

tured by DuPont, and consisted of a nickel plated 6OCo source IV 1.0 mm Jong, with a

{

(


diameter of 0.7 mm at the end of 0.76 diameter 8 m long piano wire. The source was

dropped into a 2.13 m stainless steel tube of 1.1 mm outer diameter and 0.8 inner

diametcr which had had its end welded closed. The wire was inserted and pushed

thp source to the closcd end. The other end was hard soldered to the piano wire.

The lcngth of the tube meant that the joint did not have to enter the elbow region

of the brass tubes. Both wires were kinked in the course of scanning. An attempt

to repair one was made by cutting off the wire 60 cm from the tip and joining it to

another wire v'ith a short soldered sleeve. This ",as not successful as the repaired

wire eventually broke just ab ove the sleeve while in a guide tube. The source wire

used for outside scanning was made from the second damaged wire. The tip of the

old wire was eut off, with the eut end soldered fiat, and dropped down a 1.5 mm

outer diameter stainless steel tube approximately 3 fi long. The end of the tube had

previo\1sly been closed by soldering to it a plug of hardened carbon steel. A 1.2 mm

diametcr piano wire 5 m long was inserted into the other end and pushed as far as

possible to bring the source close to the closed tip of the tube. The piano wire was

then soldercd to the end of the tube. Because the source was disjoint from the piano

wire, a weak point in the steel tube was created where the source and piano wire met.

This \Vas done deliberately because it ensures that if the tube were to break, it wou Id

break away from the tip and not damage the SOltr':ë.

The problem with this design stemmed from the nonuniformity of the source wire

due to the soldered joint conneding the steel tube to the piano wire. It was often

, difficult pushing t.he wire into the tubes past the elbow. The reason for this is that

because the piano wire is more flexible than the steel tube, it bunches up, increasing

the friction with the brass tube. This could be solved by using a longer length steel

tube covering the entire piano wire.

The source's activity of 2 mCi was chosen so that the signal from the saCo source

would be roughly the same as that from the uranium noise.

.'

-

CHAPTER 3. 60CO SDCRCE SCANNING 42

Distance from source Calcu\ated Dose

[cm] for a 2 mCi activity 60Co source

7 5510 pSt,/ h

15 1200pSt,/h

60 75 Jl8v/h

80 42J18vlh

100 271lSvlh

120 161LSvlh

Table 3.1: Radiation Dose from a 2 mCi 60Co source l'S distance

3.5 Safety

Of a rather major con cern are the safety aspects of using a strong radioactive source.

The tip of the source wire was kept in a lead 'pig' and the wire was fixed with a

clamp when not in use. Even then the radiation dose at the sUl'face of the pig was

100 pSv /hr requiring the pig to be stored in a large cabinet and surrounded by lead

bricks. Table 3.1 shows the calculated radiation dose for a 2 mCi source at various

distances. As a comparison, the radiation dose from the uranillm of a module j-. ,12

J1SV /hr at the surface. The average dose from natural background radiat.lOn i~ ahout

2 mSv per yeal'. Wc see that at a distance of slightly less than a metr(>, the radiation

from the source is appl'oximately the same as that from the surface of a module, TIl('

outside scanning procedure was rather safe. The control hut, from whc/'(' tllf' l'Un

control was operated, was several metres frorn the module stand. Clobe cxposure to

the source could only occur if one was not careful during the loading and unloading

of the source, or if for sorne reason the source was unablc to enter the tubes and had

to be manually aided.

«

(


3.6 Preparation

The scanned modules were set up in a preparation stand with the faces of the towers

facing upwards wherc they \Vere first checked for light tightness and dcfcctive PM

tubŒ. The high voltage to the Cockcroft-WaIton bases of the PM tubes was provided

by 16 channel CAMAC controllers whieh also read the monitoring voltage. Each

contl'ollel' includcd a 12 bit DAC to suppl y the voltage and a 12 bit AOC to read the

monitoring voltage. These photomultipli.~r h'6h voltages were set sueh t.hat signal

from the uranium noise \Vou Id be the same for similar sections of every tower. This

was don<.> through autotrimming. By knowing certain parameters of the PM's base it

is possible to iteratively reach a desired gain. The correct high voltage for a requested

integl'ator va.lue from UNO can be found after only a few iterations. The typical HV

on a PM \Vas about 1100 V. The UND values for the FCAL EMC PMs were set at

one fifth of the HAC PMs. The HACO to\Vers, with four times the surface are a of

the EMC towers hac! their UNO values set at four times the EMC values. The PMs

were left alont> for a few hours to stabilize their signaIs. After every few towers had

been scanned a new UND fun \Vas taken. The UNO values recorded were actually tre

average over 500 mf'asurements. There was effectively no difference between inside

scanning and outside scanning as fa'. as the driver and data acquisilion systems were

concerned. Once the driver was in place the procedure of pushing in and pulling out

t.he source wires was common ta both methods.

3.7 Scan sequence

Once the driver was in place and the source wire loaded, scanning could start. The

operator entered the tower(s) and tube type(s) to be scanned to run contr~l program.

The program then looked up in a database the correct tube number and tube length.

This wou Id he relayed to a microprocessor which wou Id handle from then on the

actual movements of the source. The driver would then position the source over the

elltrancf' to the correct tube. The source wire was pushed into the tube in 5 stages[26]:

1

t

CHAPTER 3. 60CO SOURCE SCANN:NG 44

stage 1: The source is pushed down 60 mm at its sIowcst specd of 1..1 mm/s. This

ensures that the source movcd safdy I,hrough the COllllf.'ctiOI1S inlo tll(' falJout.

stage 2: The source travels at 60 mm/s IIntil it l'eaches a distallce tubt' kllgth - 2000

mm (hefol'e the clbow in FeAL).

stage 3: The source moves a distance of 400 mm around the bcnd at 20 mm/s.

stage 4: The source continues on at 40 mm/s until it rcaches about. .50 mm from t II('

end.

stage 5: The source goes at 4.4 mm/s until it reaches the tube end. At this p()inl

l',he run starts.

The microprocessor then informs a MVME-135 computer using the OS-9 op('rat.ing

system (this comput,~r will subsequently he referrcd simply as the OS-9) that. th(' l'un

can start and proceeds to draw the wire out at its slowcst ~pc('d. t-.lcclIlwhilt' tl\(' OS-

9 occupies itself with taking data. It receives only status information aud positioll

updates From the microprocessor. Once a run is over and the source \Vire is back in a

standard 'home position" the microprocessor ,,,aits for further in!>truct ions fI' Jm the

OS-9.

3.8 The Driver

The source driver is driven by a stepping motor from the finn SIGMA and conlrolled

by a Motorola microprocessor 68HCll with a 6800/6801 cpu. The microprocessor

runs a program written in FORTH downloaded from an OS-9 MVME-l:35 computer.

They communicate via a RS-232 seriaI line. The microproccssor commlliliral(>s wilh

the stepping motor through a Superior Electric 430-PT mùtor translator wllich in t UfII

generated the correct drive signaIs to operate the stepping motof. The> Tll.,ltJI tran:-.la

tur requires four TTL Icvels from the microprocessor, controlIinl!; ..,I('p mod(' (half/full),

winding direction (clockwise/counterclockwisc), motor power (ocl/off), and I)I'o\'iding

(

(

CIlAPTER 3. 60CQ SOURCE SCANNING 45

pulses signalling a step. The controllayout is shown in Figure 3.6 The stepping motor

can be set at either full step or half step mode, offering a choice of either 200 full or

400 Italf steps per revolution of the motor shaft. Movement of the driver is controlled

by a val iable speed interrupt generated square wave from the microprocessor. Two

counters are set in the microprocessor with the second having a value WIDTH greater

than the first, where WIDTH is the width of the desired pulse in clock cycles. When

the micl'Oproccsf>or clock equals the value of the first counter, an interrupt occurs,

setting an output port to the motor translator high. The counter is then advanced

by a value PERlOn where PERlon clock cycles is the desired period of the square

wave. \\'IDTH cycles later a second interrupt occurs, toggling the output port low

and tlI(' second coullter is also advanced by PERlOn. The motor driver shaft turns

one of a pair of rubber rollers pushing or pulling the source wire. With a rubber roller

diamcter of 2 cm, each half step mo\'es the source wire 0.16 mm. The slowest velocity

was obtaill~d in half step mode, providing about 4.4 ±3% mm/s. The fastest velocity

was obtained in full step mode, about 80 mm/s. The source driver could push the

wire with a force of about 10 N. At lower velocities the motor provided more torque.

The layout of the driver is shown in Figures 3.7 and 3.8 The position of the source

wire is followed by a REX-32 shaft encoder rotating with a second pair of rollers turned

by the moving source wire. The encoder gives 400 pulses per revolution. They are

sent to a 12-bit HCTL-2000 decoder chip on the microprocessor board. The decoder

multiplies the nurnber of pulses from the encoder by four, giving 1600 counts per

revolution and writes it. into memory for quick readout by the microprocessoJ'. The

micl'oprocessor reads this value every 300 ms, check kg if an overflow had occul'red or

if the source ""ire has stopped moving. The jitter of the decoder is less than 20 counts.

Using 1 cm radius rollers, the count. to distance conversion factor is 255.6 counts per

cm. If the distance travelled since the last readout is not enough (about 1 mm at

slow speed) the driver will continue trying for a few more readings before stopping,

believing that either an obstacle or the end of the tube had been reached. The run

could only start if the the source stopped no more than a few centimeters short from


the requested distance to travel down the tube. The decoder was calibratt'tl for 1

cm radius rollers moving the 1.1 mm wire used for the insid(' ,>canning. Th(, sOllrct"'

wire for the outside scanning had instead il thickness of 1.:) mm. DIH' to tht' t'xlra

compression the rollers used for outside scanning had an ('ff('di\'(' radius of ouly 9.8

mm. This introduced a scale error of about 2% in the distances rccorot'o during

outside scanning.

After scanning a tube the source wire was pulled out to 'home position' with the

tip positioned a little before the rollers. A slottcd optical switch locatt'd bC1H'ath

the rollers would tire if it could not detect the presence of the SO\lr(t' win', stopping

the motor. Without this optical switch the source could 1)(' pullc-d Ollt, so it wa.<;

important that the threshold of the switch be propcrly adjus!'('d. Tht' 'l'TL ~igi1al

from the switch coulel be monitored externally by a voltmeter.

Once in the horr~e position, the driver would change to t.he next tube. The driver

assembly moving the source wire was mounted on a dial which was t h,> largcst of

thl'ee intermeshing bear wheels. Another stepping motor, idcntical to t.he first, ami

controlled in the same manner by the microprocessor l turned the srnallebt gear wlH'd.

The gear ratios were 154:36:36. The dial lies just above the base plate of the driver

assembly. The base plate has a ring of 100 holes equally spaced. The hales are labeled

from -50 to 50 with 0 being home position for the dial and -50 and 50 f('I)}('st'nting

the same hole. These holes lead to the short tubes of the extender. Not al\ hol(·s

corresponded to a brass tube in the module, but a map of the IlOle assignmcnts was

kept on the OS-9. The dial driver positioned the source driver over the appropriate

tube. 16 half steps separated neighbouring tubes. Before changing to the tube for

the next run the dial would first turn to its home position. Attachf'd to the source

driver's assembly was a metal finger which would tl'igger a st'cond optical switch on

the base plate when the dial was at the dial's home position. Then the dial would

rotate to the next hole.

•

{

1

CIlAPTER 3. 60CO SOURCE SCANNING 47

3.9 The outside driver

The ùriver used in the outside scanning was modified from the original driver used

for inside scallning. The entire driver assembly was mounted about 40 cm ab ove a

platfol m equipped with two l'allers with wheels on the end. The wheels roUed along

the sicle of the module as the driver was moved from tower to tower. This gave the

platforrn extra stability against tipping. Sitting on the platform underneath the dial's

home position was a lead pig used for storing the source while the platform was moved

to a new tower. Access to the pig was provided hy a hrass tube connecting to tube O.

Only nine other tube holes in the bottom plate of the driver assembly wcre used, four

per si de for scanning and one extra for loading the source wire. These holes lead to

the brass tubes projecting from the aluminum wings. The tube database on the OS-9

was modified hefore every tower 50 that the run control program would use only these

tulws to scan every tower. The free end of the source wire \Vas guided through a 30

cm vertical brass tube immediately above the rollers and threaded through a high

aluminum profile abo\'c the driver to help prevcnt sharp henn.s occurring in the wire

near the source driver. A camera, mounted on this profile, was connected to a TV

scrcen in the control hut to allow the operator to remotely monitor the operation of

the source driver.

3.10 Data Àcquisition

The readout electronics used in the 6OCO scans were based on those that will be

uscd for the slow control during the ZEUS experiment. The signaIs from the PMs

were colleded by special integrators designed specifically for the ooCo scans \Vith an

integration time of about 24 ms. During this time they would each receive 4 - 5 . 105

pulses. The integrators were read by 12-bit multiplexer analog to digital convertors

(MUX-ADe) in a VME crate. Each MUX-ADe card has an ADe which read out

96 independcnt channels with a total readout time of at least 50 ms. In practice the

time delay between consecutive readouts was 60 to 90 ms, due to the time needed

-

------------------------......... CHA.PTER ,1. 60CO SOURCE SCAX~\'ING

ta transfer the data ta the OS-go The range of the MUX-ADC was ± 5 V. Thlls

one ADC channel was about 2.5 mV. The ll~O values \\'t.'l'(, St.'t al 500 IllV for t:it'

FeAL E~'IC, 1000 mV for the ReAL E~1C, 2000 mY for the I1ACQ, and :-!500 m\' fol'

the BAC sectiOllq. The 90 m!:> reil.dout rate mean! that lllt'a!>lI1'('Illl'llt., "t'l'f' tah'lI a\

11 points per second. This conesponded to 2-3 mCaSU1'Cllwllt points !wr IllllliIllt'tre.

The source position was updated in t he data aftel' evel y 10 llwa-,u r<'llll'llb. TIlt' 011'

line analysis plograrns interpolate bctwcen the l'ecorded po,->it ions 10 obtclin 1 II(' tl\\t'

position of every data measurement. The raw data could be plottcd on a Ü'l'III IIH,,1

during the run. For data taken from inside scallning. disldllce \\',IS 111('.-1"111«'<1 hum tl\('

end of the tube by the front plate. For data taken dUlillg ()lIt~id(' ~(,(1l1l1iJ\g, dl~t;lI)(,(,

was mcasured from the end of the tube just past the Idst BAC spction.

For every fun up to 12 PMs werc rcad out, 6 on e3ch sidc of the mo(lnle. ThC'

raw data were recorded in ZEBRA EXClIAt\GE format[27]. ZEBRA is a. dilta man

agement system supported by CER~. One fun in FCAL took ahout l 3 Mbytcs of

space. This data was laier concentrated 50 that only the average fH'r mm WdS k<'pl

[25][26]. In this way a run required only 180 "bytes of storage. A larg(' Fe AL module

required about a hundred runs. The OS-9 had only about 80 Mbyte., of disk spart> ~o

the raw data was transferred over ethernet to the central D ESY VAX wlwl'e it was

later concentrated and put onto magnctic tape for pt'lll1ancnt ~tOl(lgC.

The scanning time \Vas about 40-50 minutes per tower, doing 2 t.llhes on (lach sidc.

The data taking itself requÎled only 6 minutes to scan the 1600 mm of a FeAL tOWf>f.

The rest of time was needcd for the source to reach thC' end of thl' tube, to change

tubes and, in the case of outside s('annÎng, tü move the scallJlJllg apparat us to a JJ('W

tower. The large FCAL modules took about 20 hrs each to scan, Ilot indlloing the

preparation time for cabling, fixing light leaks, etc.

In scanning ail of the FeAL and ReAL modules, somc 5000 runs were taken,

including repeated runs and special measurements. The raw data filhl 40 rnagnctic

tapes (6250 bpi), the concentrated data another five. The ('nt.ire out "ide scanning

operation occurred betwcen October 1990 and April 1991 at DESY.

(

(

CHAPTER 3. 6OCO SOURCE SCANNING

1.0

0.6

0.2

1.0

0.6

0.2

0.2

0.6

60

EMISSION (normatizfCI)

-Y-7inPMMA CKamon f1 ail .

EMISSION (normalized) SCSN-38 (BOB in PS) CKamon ft al)

a)

ABSORPTION (ncrmaliz~d) --Y-7in PMMA

[0YIr\ mfasurtmentl

SPECTRAL SENSITIVITY SbRbCs CQthod~ [VALVO hand book]

1 1

350 450 550

Figure 3.1: Absorption and Emission Spectra of SCSN, Y-7, and PM cathode

49


1

Figure 3.2: View of the Source Guide Tubes and Light Guides


(

lOUrCe wI,.

-len ~

R DRIVER

l' , i " ~

E~ 1 2 3 4

HAC1

HAC2

PM'.

( Figure 3.3: Sketch 01 the outside driver scanning a module.

...,.

CHAPTER 3. 6OCO SOURCE SCANNING

•••• \~ • o •••••••••••••• or· .......... ·~J~~I~~~· .. €~ ... ~~ ..

f l

~ -... ------

steel tube -lmm

.... - - --, 1

Figure 3.4: Source wire used in inside scanning

4 Meter 3.10 Meter . . . .... ...... --... ! 1 ,/ "'" 1

i ll: I~, ... j ........... ou ••• ("' :~:;d-t;~:.~·· .... ····· t .::::::: lE' ..

.. 100 mm

•• , ,

,lo.... ~ ,/ -------....-

-3mm steel .eal

•

,

l ~l.œ~-R~-~r~r-~\ii paano ) ~! L- ___ ,

Figure 3.5: Source wire used in outside scanning

CHAPTER 3. 6(JCO SOURCE SCANNING 53

~_M_A_G_T_A_P_E __ ~I~~ ____ D_IS_K ____ ~ .r-~

DISK MUX-ADC

INTEGRATORS

HV CONTROLLER

PM TUBES

DECODER DRIVER

{ Figure 3.6: Layout of Elements in the Run Control

CHAPTER 3. MCO SOURCE SCANNING

1

.otor

. ....... ,-.

base plate

Figure 3.7: View 01 Source Driver Irom above

dial 8Otor

Figure 3.8: View 01 Source Driver Irom the side

(

Chapter 4

Monte Carlo

Il is often helpful in understanding complex systems to simulate them \Vith Monte

Carlo programs. The adual processes involved may be individually relatively simple,

but multiple consecutive interactions can quickly hecome exceedingly complicated

and not analytically solvable. In our situation, the process of shower development

is statistical in nature and lends itself naturally to Monte Carlo studies. The Monte

Carlo program EGS4[28] was used to simulate the 6OCo scans. It is a successful

program for simulating elcctl'omagnetic showers.

The user has only to define the detector's geometry and properties of its materials,

specify the initial conditions of the input particle, choose the energy cutoffs and

kc<,p t.mck of the cnergy deposition. Every region has an energy cutoff ECUT for

e!('clrons p .u peUT for photons under which they will no longer he tracked. The

particles are then assumed to deposit aIl their energy in that region. EGS4 (or its

preprocessor PEGS) calculates the interaction cross sections for the rnaterials. The

user defines rnaterials as compounds or mixtures of elements. For every material an

AE and AP must be defined. AE and AP are respectively the energy cutoffs for

electron and photon production in the material. Particles move in small steps, with

each step bringillg them either to an edge of a region or to the point of their next

interaction. Also, for low energy electrons, a step size, which is the fraction of an

electron's energy lost to ionization per step, can be set.. This gl"nerally decreases

55

. '

CHAPTER 4. MO.'\'TE CARLO

the step di~tallc(, for low energy clectrons, increasing the accllracy of their tracks .

~atUl'ally the accuracy of the simulation improvcs with the low('ring of the PIl<'rgy

cuts. CnfOltunately the computation tinw incr<'1\ses as wl'II SI'\('lal ditfl'I(,l\t cuts

were tried out and t!lC!C did not Scem more than a fl'\\ Pl'ICt'lIt difTl'I('IIC<' !}('tW('('1l

them. However the computation time varied sigllificantly. The COlllputdtion till\('

was Ilot significantly lcngthcned by using the lowest 'Berg)' cut allo\\'('d for photons;

10 keV. Belo\\ this limit the approximations for the cross sectIons blt>éÜ. <lo\\'n. The

main drain of CPU time are the electrons, which maye in small steps, depositing

ionization energy and un(l ergoing multiple scattering Th{' au! bon; of tilt:' plOgram

rccommend using a smaller than default stcp size when dCilling wlt.h thin la)f'l's. ln

the end nonstandard cntoffs were used: FeUT ::: 1.0 Me\', }l('\1'}' = .0\ ~I{'V, AE

= .711 ~lcV and AP :.::: 01 ~1eV everywhere. A step size of 5% was used for sU't'l and

scintillator. The default step size wa:; used in uranium, SlIll'(' it was a~~l\IJ1{'d t hat

most low energy clec! rons would Ilot escape the mélIlillIll ond »<'1]('1,1 ctte t hl'ough the

steel foil surrounding it. For a million ('vents it took about .1 Itol\rs ('Pl! on a VAX

8600.

In the l'l'al 60Go scans the data is takcn rollghly every .5 mm in the longitudinal

direction, z. If we were to simulate the GoGo scans in the same mannf>r, i.e. by

repositioning the source longitudiually along a tower, the étlllOlint of computer lime

required would be prohibitive. Instead, we can take advantagc· of the p .. riodicity of

the layer structure of a BAC section to find the signal re"'ponsf', R(z), of a singlp

scintillator as a function of the source distance in the longitudillal directioll from the

centre of the scintillator. We can then convolutc the respon,>e fI 0111 a ~illgl(' ~(intillator

to get the signal from an entire section. A way ta do thls is la fil~t deflll<' tl)/' g(,olll('try

as having a lalge stack of DUjscintillator (in this cast"' 120 Id)els) to avoir! havlIlg to

worry about edges. The number of layers is not irnpol tant a~ IOllg o.'> t}l(> dl~tallce

from the middle to the end is greater tItan the range of ;; III whi< h Wt:' éIlC int('r('~t('d.

Next wc run the Monte Carlo with the SOUfC(' along the stack al 8 dlfTf'f<'nt po~itions,

spaced 0.5 mm apart in the z direction, starting at 0.2.5 mm f/Olll the (f'utn' of the

( ..

CHAPTER 4. MONTE CARLO 57

middle sc:intillator (callcd the Oth layer, but actually the 60th of the 120 layers) of

the !>ta.ck lü .1 Î!') Illlll from the centre. At the eighth position the source is actnally

oppo.,it(· an III iiniulIl layel If the "ource is at a distance z from the middle of the Oth

layer, thcn il i~ a di~t,U1ce 8z - z from the middle of zth layer. This means that for a

source ::.itting at position z, the energy deposited in the 7th scintillator is R(8i - z).

If wc also use the fact that R(z) =- R( -z) then the Oth scintillator gives data for 0.2.5

- 3.7.5 mm; the 1 st scintillator covers 7.75 mm to 4.25 mm; the -lst scintillator is in

the range 8 25 mm to Il.7,) mm and so on. By combining the information frorn each

scintillator layer we can construct R(z) for any z. This function R(z) obtained does

have a slight correlation between points separated by a multiple of 4 mm because the

values were gcnerated by the same Monte Carlo fun, but it should be removable by

smoothing algorithms.

Once we have R(:) wc can find S( z), the SUffi of the l'esponses of eighty layers, by 79

S(z) = Ec,R(z - 8l),

whcre CI i., a measure of the efficiency of scintillator layer i and z = 0 mm is the

location of the Oth fcintillator.

A simplified but nevertheless detailed geometry of a HAC section was used. This

is shown in l'igure 4.1.

The tower was assumed ta be large stack of DU fscintillator tiles. The 60Co source

\Vas treated as a point source emitting 1.17 and 1.33 MeV photons with equal prob

abitity in a random direction in the hemisphere towards the tower stack. The source

was placcd outside the module as in an outside scan and 5 cm away in the laterai

direction From the centre of the tower.

The raw shape of the source response is shown in Figure 4.2. The error bars ar~

purely statistical and have no bearing on how weIl this shape matches the l'cal results.

Actually, special measurements \Vere done on severai towers with the 60CO source.

ln one su rh measurement, the strap was removed and the edges of al! but one sein

tillatol' layer wcre co\'ered. A 60Co scan was then done normally. Figure 4.3 shows

the l'cal response of a single scintillator. The fundion obtained from th~ Monte Carlo


1

)59 more 1 l 1 1 1 1 1

loyers 1 200 -0.65

AIR

2.6 SCINTILLATOR 0

:o-6C ,-... Cl> sourCE 065 -..... AIR QJ (f) U1

\ 0 0.4 (f) V1 u

FE cr: -l -l + cr ~ ~ 3: < c-

o ~ ..... (f)

'-'" 3.3 URANIUM w w 0:: 1.0 ~ u... <

2.0

0.4 FE 2.0

0.5 z l60 more 0.4 06

3.0

Lx loyers

-

Figure 4.!: Layer Structure used in Monte Carlo Simulation. Distances are in mm.

'600

CHAPTER 4. MONTE CARLO

1

j '200 r

~

~ 800 l

!

le"",,)

t

i \ 1 r r

o

\

\

59

1 . 1

l j ~

, .., . " '~'lij

~o ,~ 100

Figure 4.2: Monte Carlo of the Response function, R( z), of a Single Scintillator,

before and after smoothing

, r · 7 _

, r •

~ -t . ~ ·

3 ':. · ~

1 2 ~

.,~

r 1

,,;

~ ,1 . , ,

\

\ .. \

... '\

~ .00 U~ ,~ t'~ 1000 toa=- 'ose ton

-'-'1-1 "'''''yl. IIIL" ru. 2~~ CODQIt ,ft To •• t 12 [WC, /O'C2l ,... l'lAC'.

Figure 4.3: Real Response of a Single Scintillator

-


is slightly nalrower thdn the real response but othel wise reproduct's tlH' !-Ih"lH' [.urly

weIl.

After doing a S poillt filter on the law shape [rom th(' ~!ontl' (';\110 •• \I\t! SlI\I.lut hing , the distribution wilh the HBOOl\ loutine llS~100F Wl' Cdn appl) tht' ,d,u\'t' 'fOrI Il Il 1.\

to get the shape of a 60Co scan from a 'good' BAC ~('ction (s('t' Fig,I1r(> .1. 1) The l'\

point filter averages over the eight neart':,t points. The filt('ring alld Sllloot hlIIg wlllt'h

decreases the correlation between points c;eparated by 8 IllTll also ~igllificalltly ~muot hs

the peaks and valleys rcpresenting scintillator and DU layers after (,o1\vollll iOB. This

plot of the H:\C section and the succecding oneS \\ere ill faet llIade by takillg thn'('

Monte Carlo runs to find the response functions for scintillat.or!'l proj('( ting floll\ t II<'

uranium by 0,0.6 and 1.2 mm. Then in the convolution, 80% of IIIC' !'Icillt.illalols wert'

assumed to be projecting out by the nominal distance of O.G mlll 'l'Il(' rl'lJlailllllg :.?O(,:{,

were assumed to be pu shed in or out by another O.G mm and Il:-.(·<1 t ht' <lpplOl'll"t('

response function. The redl sciutillatoIs are exp('c1('c! .,hghtly :"llIftt·d hl'( <Ill"!' of t II!'

culouts in the scintilldtor for the spacers ale slightly Ictrg,cr t h<lll tilt' "!lM!')' ~JZ('.

This makes the shapcs of the individ ua} scintillatOl peaks ~e{'m les:" 1 <'gllictr éllld mort>

realistic. In addition, we assigned randornly to the coeffici(>llt c, of R, (z) a valllt'

between .95 and 1.05 to reflect minor uncorrelatcd variancc5 in scilltillator thickll('s.,

or WLS uniforrlllty.

We can sirnulate what potential errors would look lik(' by adju!-II ing l'a tü mimic t 11(>

effects the errors would have on the the light output of the sClllttllators. For ('xalllplc,

Figures 4.4 - 4.7 show plots with a 50% shadowed 5cintillator, bad stacking with !-IomC'

scintillators pushed out, bad stacking wilh sorne scintillators pu<;hed in, and a lill('ar

dip in WLS response of 20% in the first 20 laye' s.

With the Monte Carlo we can investigate in detail somc of the factols which affect

the 6flCO scans which cannot be easily discovered expcrimcnt ally. For ('xalllpl(" the'

average depth where energy is deposited in the scintillator and con~cqll('ntly, whcrc

light is produced, is rcvealed through the Monte Carlo to be about }(J mm. Figllre 4.9

shows the deposition of enelgy sumrned over ail the scintillators in the x-y plane wlth

(

c

CHAP'fER 4. MONTE CARLO

~ ii > ; 1. 0

P '" Jo 0 r

11000

7000

6000

5000

JOOO

1

1

1

::~ o 0 'OO:-~2=00~-~J~00~~.00~--~~00·~--~600~--7~00~~

No, mOl ti&C Sect1ot'l Sourte t)OlltlO~ lmm)

Figure 4.4: Convoluted Response Fundion Representing Signal From BAC

i ~ l. 0

l 0 r

11000

7000

6000

!IOOO

4000

JOOO

2000

1000

thodo •• d ICInt

100 200 .00 100 Soure. _,\_ (mm)

Figure 4.5: Atonte Carlo HAC section with one scintillator 50% shadowed

61

1

t

CHAPTER 4. MONTE CARLO 62 . ! ---_ .... -- -. ,

9'lQ() i 1 , "t'd"CeG '*lS '."0 &

1~Jr(r~Itf~I(fIIIfhM~ Z; !OOQ : 5 , " 1 <

'" 'aoe : ;, e • Il. ,

J , 6COC ;

, ,

5000 l , \

1 '000 t

1 1 ,

lOOO r 1

2000 t 1000 r

o ~---- ~--~-"" o 100 20C Joo '00 '>00 600 100 800

ShOl"tldt" Soufte POSI\'Of"I (mm)

Figure 4.6: Monte Carlo of HAC sectIOn wlth damaged WL8. The first 20 layfr!> hal'f

a reduction in Z'ght output of 20% for the first laye l', den'easzng lmear/y to 0% for the

tWf.71ty-ji.rst layer.

JOoo

.00 -~- -600-- 700 100 200 100

Figure 4.7: Monte Carlo of HAC section with 120 layers pushed in by .6 mm


1 9000 f , 1

~ . ï; ecoo ~ ;; 1 ~ ~ l 7000 ' • 1

t' t 6000 t ~ooo

-000

)(lOC

'00 200 JOO 100 !>OO 600 100 100 50"'0. CIO"\'O" (mm)

Figure 4.8: Monte Carlo of /lAC section 1J.'lth eo layers pushed out .6 mm

63

a 6OCO source at the origin. From the figure it is scen that not only does the 6OCO

source have a limited penetration depth in the scintillators, it also illuminates on)y a

!lmall region along the edge of the scintillator. This implies that the 60Co scanning

can be very sensitive to edge effects and very local inhomogeneities which rnight not

affect most particle showers because most showers are expected to occur away from

the edges. Cnly part of a shower can occur in the edge regions because the cracks do

not point towards the interaction region.


l 120

j 100

1 80

60

40

20

20

64

EDEP IN SCINT X-Y

Figure 4.9: X- y distribution of energy deposited in scÎntillator with the source posi

tioned at 8 mm away, as in outside scanning.

•

('

Chapter 5

Results of 60Co Scans

D('(or· ;oing into an analysis of the 60Co scans the conventions used in describing a

l'un should be dpfilled. First, we distinguish betwecn the two sides of a module by

lert and right, as defll1cd from the perspective of someone sitting in the interaction

region and facing the module. This also holds for the source driver. The guide tube

labelled EMCI jEMC2 runs between EMCI and EMC2 in FCAL and over EMCl in

ReAL. The tube labelled EMC3jEMC4 lies above the gap between the EMC3 and

EMC4 WLS in FeAL and is over EMC2 in RCAL. The tubes used to scan the HACO

towcrs are called either HACOI or HAC02 and were positioned respectively on the

lowcr or uppcr side of the tower.

5.1 6°00 Signal

The signal from a calorimeter section undergoing a 6OCO scan can be divided into

three parts:

• Uranium noise signal

• 6ne...: 0 signal from scintiIIator

• 60C 0 signal from \VLS (Cerenkov light)

65

... ..

CHAPTER 5. RESULTS OF saCQ SCANS

3000

2~00

2000

1~00

1000

1

1 1

1 1

\ u~o

~~Ot. L~~,~·~·~·~·I~.~,_.~I~.~,~~. ~I~.~~~~~~ 200 .00 600 100 '000 '200 "00 '600

,o",r(, E)Q'lttQf'l (m~)

Figure 5.1: Raw signal from PMs readtng out both sldes of a HAC sectlOTl

66

An activity of 2 mCi gives a 60Co signaljUI'\O ratIo during outside scanning of about

0.7 (0.5) in the HAC l!{'ctions and 1.6 (1.2) in the E~tC s{'ctions with the source on

the same (opposite) side of the PM. The signal/UNO ratio for the inside scans were

about 15% higher. The Cerenkov light is only seen when the soune is on the sarne

si de of the PM tube being read out. In order to compare difTerent towers we use,

instead of the raw 60CO signal, the signal/UND ratio = (raw signal - UNO)jUNO.

Sample plots of the signaljUNO ratio as a function of source position for norma.l

sections are shown in Figures 5.2 - 5.5. These plots were ail ta.ken using the outside

scanning procedure.

Each of the peaks corresponds to one scintillator. The little dips occur when the

source is directly opposite an uranium plate. The EMC and HACQ sections in FCAL

(ReAL) have two (one) large valleys corresponding to the positions of the silicon

gaps. The signal Îs reduced there because many of the garnmas escape out the oLher

end of the gap .

(

-

CHAPTER 5. RESULTS OF 6OCO SCANS

! t 07 r • t'

06

O'

... ~ ~r ____ ~,~~ ____ ~ ____ ~,~~,~,~. ~5oo '00 800 900 '000 "00 1200 'J(,() 1400

IOurte DO"tion (mm)

Figure 5.2: Signais from PMs reading out both sldes of a normal HAC section

~ '-16

l i3 1 4

lOll't t 01'1 ,Dm, 'Ide 01 reoOou\

, 2

...... SOurC' on OOPO,lt. liGe 01 'lIOcloul " • ,

01

06

"

•

0: r~' ~'~'~'~'~~'~'~'~'~'_'~' ~,~~!~ __ ~,~ __ .~~~~~~ • J20 , 360 "00 14.0 ' 4!0 '520 '~60 '600

67

Figure 5.3: Signais Irom PMs reading out both sides 01 a normal EMC section an

FeAL

CHAPTER 5. RESULTS OF eoCO SCANS

011 1-, f ,

:

ID'J'CI 0'" oppas.t, s.a. ... ... ~

os "'.-odoul

68

Figure 5.4: S,gnals from PMs reading out both sides of a normal HACO section ln

FeAL

, 2

l

.~ ,; 1 . ,- . ~

08 / . .o~'ce 0'" OPPo"I, 'IC" 01 '.oaout

06 / . ,\ 'i., 9Q I>

~

Ot / l '

02

o ,-"-,~~~~~..........,~ ....... ,",,"::.L....,,, l, •. 1., • l , 680 720 760 eoo IItO 110 910

'0''''. po.ttlon (mm)

Figure 5.5: Signais from PMs J'fading out 60th sides of a normal EMC in ReA L

., l

f

CHAPTER 5. RESULTS OF 6tJCO SCANS 69

~ 01

l o

l' O~

Ol

02

01

o 0 1 00 200 lOO .00 ~OO 600 lOure, po •• t.Or'l {m~}

Figure 5.6: Scans of Module CDN~, Tower 1.4 HAC! wlth the inside and outside

methods

5.2 Reproducibility

Before trying to discover assembly faults from the analysis of the data from the

60Co scans it should be shown that the results from the 6OCO scans are reliable and

reproducible, to dear douMs on whether the inhomogeneities found are in fact caused

by inhomogeneities inherent to the tower, or are mere artifacts from the scanning

procedure. Figure 5.6 compares the results t'rom the inside scanning and the outside

scanning of a. HAC section. Despite the decrease in signal in the outside scanning, it

is clear that both methods reveal the same gross structure.

As t.est of the position accuracy of the source and its sensitivity to the the layer

structure of the calorimeter we can determine the length of the sensitive layers of the

HAC and EMC sections. This can he done by looking for the difference hetween the

positions at which the gradient of the 6OCo signal is at a minimum vI a maximum.

Presumably this should occur when the source i5 just before the first or just behind the

last scintillator. By comparing the results from inside scanning and outside scanning

1

CHAPTER 5. RESULTS OF soCQ SCANS 70

la

'0

Figure 5.7: Width of HA Cl sechons from inslde and oulsidt' scanning

\\le can also clearly see in Figure 5.7 the diffcrence the thicker source wire makcs to

the scale as cliscussed in Section 37.

The quality of the scans decreased somewhat in gomg from inside to outside

scanning. The layer structure is still visible with outside scanning, but the scintillators

at the borders of a section are less pronoun.ed. The Fe AL E~lC sections were fairly

sensitive to the precise location of the guide tube because the tube!> run a)ong the

edge. The shape of the plot can be significantly distortcd if the source runs over part

of a neignbouring tile. This does, however, make the peaks more proJl'C'\Allced bl~cause

the uranium acts as a collimator for the scintillator layer c10sest to the source. The

priee paid is a decrea..<;cd signal/noise ratio and a greatcr scnsitivity to smal1 variationR

in the source position. The inside scanning tubes are placed cxactly bctween the EMC

WLS, whereas there were sever al millimetres lceway with the alumil1uTn profiles used

in the outside scanning. The inside guide tubes have the best of both' good r('solution

of the border scintillators and a weIl defined position pro\ iding a consistent shape of

the scan data. Ali other scintillator t,les had the source tubes over the tiles, away

from the edge, so this \Vas not a problem for them.

c CHAPTER 5. RESUI.TS OF 6()CO SCANS 71

5.3 Faults

Ali the modules were scanned by the outside method. The results from the scans

were plottcd and examined individually by eye. The 6OCO scans provide a visual

impression of the quality of the section examined. Ml\ny types of potential faults can

he characterizcd by correlated bumps and dips from diffcrent plots macle with different

somcc positions or l'l'admIt sides. Recognizing faults was akin to pattern recognition,

looking for Jclatiollships between the different scans. Naturally the human brain

cxcc!" al such a task. The time to go throagh the plots of a module was a few hours,

dcpcnding much on the cxperience of the examiner. During the scanning of a tower,

the operator could in principlf' al50 go through the results of a previous module.

Although computer pl'ograms[25][29] were dcveloped to assist the analysis they were

too undcpendable to cope alone with the wide var:cty of possible faults. They were

used, however, to check the length of a section to see if a layer was 'missing'.

Faults arc charactcrized by how the response varied with source position and

ou which ~ide was the readout. In general, assernbly faults occurring in the EMC

section were of t.he grcatest concern. Because the EMC section has only 26 layers, I.m

afTected scintillator therein contributes 4% of the signal, whereas in the HAC sections,

a single ~cintillator comprises only 1 % of the active volumf."'. In addition, EM showers

are almost totally contalllcd within the EMC section As a consequence repairs were

/llostly do ne on the EMC sections and S0111e faults in the HAC sections were left

unrepaired due to a lack of time and manpower.

5.3.1 Bad Stacking

If the spacers were not properly positioned during stacking, the scintillators could

end up shifted from their design position. This is quite noticeable when it OCClUS to

a group of adjacent scintillators and is called bad stacking. The EMC2 and EMC3

tiles of FeAL were more prone ta shift because they are not held in pLiee by spacers.

Sometimcs the scintillator tile could be pushed back into place, but if the spacers

1

0.4

03

02


\ \ 1

.\ 1 =-

. , -:

i

72

\" -:c ecc iC; • ~CC cc ::C !CC '::

IOWIru paa\\oli Im._

.... 00 .... "1,3 n"t'I ,2C3 "DC' n ~Ow.,. '! ~\lC~ '("'C"'l, ;J'A .. .... CH.

Figure 5.8: Bad stacking

were too far out of position, this could not be done. The had stacking effects are seen

by an increase (decrease) in signal in the region where the scintillators are pushed out

(in) on the side where the source is. The change in signal is a result of the change of

scintillator volume directly exposed to the 6OCo source. Figure 5.b shows an example

of bad stacking.

This is generally not considered to je a serious problem hecause the increased or

decreased scintillator exposure is amplified by the source's position at the si de of the

stack. No effect should occur for small shifts «0.5 mm) durinô a real shower where

the particles are travelling transversely to the face of the <:ri'ltillator unless they came

straight down the edge of the stack. For larger scintillator shifts, it is possible that

one end of scintillator will be shadowed by the uranium plate such that sorne of the

light exiting the edge will be prevented from entering the WLS (se(> Figure 5.9).

L

CHAPTER 5. RESULTS OF &aGO SCANS 73

... o..

u 0.1

o.. o..

o 0 100 )00 lCID 0lIl) _ _ '00 eoo °O~~1~00~~2oo~~lCID~~0III)~~_~~_~~_~~_ -_1-1

.,....,.. CC*l."Wt 10" c""lfIl .... CMC I/[IIICZ. ....-..cA -,..-1-1

,.,..&HiI co.J Mt 101. c",11f11..., • (1IC.l/EMC"" ,..- ""';211

Figure 5.9: Shifted Scinhllator

5.3.2 Bent WLS/Optical Contact

An interesting effect related to bad stacking occurred in module NL14. There the

scans showed a significant increase in the response of the third scintillator on the right

si de independent of source position. Figure 5.10 shows the results of two scans, taken

from the left and right sides. This effect was seen to varying degrees in over half the

towers of NL14. Upon visual examination of the affected towers, three observations

were noted. The third layer of scintillator had been shifted to the right. The nylon

fishing line which normally separates the WLS and the scintiHator edge had dug a

groove into the edge of the scintillator. The WLS had a slight bend around the

outcropping third scintillator, as it is constrained by the strap to fit against the edge

of the front plate.

This suggests two possible explanations for the increased signal. One possibility

is that the WLS and the projecting sCÎntillator are in partial optical contact. This

would allow light from the scintillator to enter the WLS directly, without any reBection

from the surface. However, this might not be the correct explanation because optical

-

CHAPTER 5. RESULTS OF etJCO SCANS 74

!..l . l i

, ,

r '2 ...

. , ... t ~

oe ~

~ 06 -

~ t : ~ ,

0' - 1 ~ 1

02 r r

\ , ,

.. 1 1 ~ --~6e~0~~7~2-0--~'-6C----!-C-O---!-'~C--~8~èo~--9~~70~

~ • 2 r

f 011

06

02

.. WU p..'ueD mm 1

'T'Odwll ""Il,. "\.+" ']3~ ~ C;:>OOlt ,,"0.,' i ( ... C: L Cl'W- [WC2l

1 ,.,

( 1

1

1

o ~~58~O~~7~20~~776~O~~8~CO~~e~.7o--~88~0~~9~2~0~ ~pc.uo.l_)

" P .. - [IoIÇ2~

! .-

~------------------

cet-

J \ 1

, ~. 1-

( ,

V '----"--~----:~-~~-.l-.... ....&...- _____ ~_ ........ J __ _

SIlC 7~: 160 800 S'O sec ~~c .'OIr' J""II,,'n mlln

'T'Iod",le "l," ~"" '.3JI cODOn ln To •• ' 9 (liIe; .. p ..... E ... ~;'R

ç------ ---~-- -- - ----- ~--- --

6 -

• -· : --· •

Cil -• , r

C 6 r

!'

,

, \

l \

\ 1 \

,. ~ J ':L L ., ,.

\ ~,

. 1.

760 ,1

aoo , , ~. 1

880 1110 611C 720 ""(1 ptat_ j •• l

"'00"'" n'" '"" BJ4 C~bol1 ," To •• , 9 (WC2 ~ P\oI- EwC2"

'" \ 1

Figure 5.10: Increased Response from Shifted Scintillator in NL14

-

CllAPTER.5 RESULTS OF 60CO SCANS 75

contact o\'er a ~ul'face is difficult to achievc by pICSSUIe alone. Tbe other suggestion

advanced was that. bccause the \\'LS is bent around the third scintillator, hght Ieaving

the ~cilltill(ltor at c,;harp allgles hits th(' WLS at an angle clo~('r to the normal than

ul>ual. l'hi., would d('( l'case the amount reflected off the surface and incr<>ase the light

yield[30].

The (('ason this effect is not seen for other badly stacked scintillators further down

III the stack IS that the third scintillator juts out near where the v\'LS is forced to

COll tact tll(' front plate. This placed a lot of pressure in the region which would not

IJavc occufl'(·d fUI t 11('r dOWIl the ~tack.

As It tUrIlcd out, N 114 was one of t he modules that went t hrollgh beam testing

at CEHN. No incr('af,cd signal response from the right side was seen. This is not

slI'lHising bt'utust' the fradion of encIgy d<>poc,;ited in the tlmd layer by a 15 CeV

el<'( lIOn is not VCI)' different from the fraction that the layer lOntributes ta UNO.

Whc\tc\'('r the cause may have been, the problem cou!d not be fixed in the limited

time and facilitics available in the DESY preparation hall and the module \Vas later

installed without modifications.

5.3.3 WLS problems

Wa\'elcngth shiftcr problems should be detectable Crom 60Co scans flom bath sides,

and be seen in the signaIs from PM tubes on only one si de. They were relatively

cas)' ta l'epail, lequiring the rcmoval and repair or replacement of the affccted WLS

cas.,ette This was done for ail problcms occurring in the EMC section and many in

the IIAC :,('ctions.

During the assembly of the WLS cassettes it was possible to put them together

Încorrectly such that after thcy were mounted, one or more sections ".lOuId not be

complt>tely rt'ad out. It was also possible for cassettes to be incorrectly mounted

such that the \VLS for evcry section would not recci\'e light from a scintillator layer

(usually the first). For the EMC section, this would usually be quite visible from the

scans (sec Figure 5.11).

-


~ 0 7 ~ 'i- r ." ;

06 r-.

01 ....

\ \

\ '

~

o . , • ~20 • l60 • -00 'UO olle '~;c '~6C • 600

"""'P<*''''l_l tTlOdlolt, :ONJ 'un l'.S. coOolt ft To.er • "'ACCt ... ~1wI- ""CCRo

76

Figure 5.11: Shzfted EMC WLS. Note the sharp drop where the first scinilllator layer

should be mdicattng that the first layer ss not seen

Shifted WLS

For this reason it is important to know the width of the sections. If the width is 8

mm smaller than expected, then it is a clear indication of a shifted WLS. Similarly,

the distance between the beginning and end of neighbouring sections can also be

used to detect shifted wavelength shifters. ldeally, for example, the beginning of the

HACI section as seen by 6OCo scans should also be where the EMC section ends. An

alternative method attempted was to count the sC'intillator peaks. A fourier transform

of the data Wa15 taken and the low and very high frequency components were removed.

Since the peaks should occur every 8 mm, only the frequencies in this region werc kept.

After undoing the transform with the remaining frequencies, the signal remaining had

very pronounced peaks corresponding to the scintillator positions. This made it easier

to count automatically with a program[29}.

(

(

CHAPTER.5. RESULTS OF .,co SCANS 77

14~----------------------------~

12

o.

01

04

02

o 1320 ' . ;3'.0 ' ';~ ';!o • Il!!, !.!

1~1O

mocIult CON4 run 1967 __ ,\ '" To_ .. 17 EWC1/EWC2R PW- EWC2l

Figure 5.12: Sticking back reflector

Reftector Problems

Refledor problems range from a badly matching pattern which does not compensate

enough to make the WLS uniform, to a sticking back reflector. Sticking back reflector

problems could occur anywhere in the stack. The back reflectors are held in place

at various points along the WLS. Occasionally they stick to it and total internaI

reflection is lost, reducing the light output. Figure 5.12 shows the result of a sticking

back reflector occuring near the front of an EMC section. There is a steep drop in

the response of the forward Iayers.

5.3.4 Light Path Obstructions

The presence of foreign material in the light path was easily detected and could easily

be repaired by sirnply opening the tower and removing the offending piece of paper

or plastic.

1

CHAPTER 5. RESVLTS OF 6OCO SCANS 78

. :;: -

<8C 6CO JOiIHt pat.lI00 m.l

""QCuc .... L2 '",,, '5 CO~O\\ 1" "o.c· ! (wC.!, ~ .... C.," P-..- : ... C,"!P

Figure 5.13: Shadowed Scinttllator

Shadowed Scintillator

As it had been shown in the Monte Carlo resdts, the sbadowing of a significant

fraction of a scintillator tile in the middle of a stack is clearly visible. The black

Tedlar sleeve was in sorne insta.nces shifted over 50 that sorne of it went pa,r,t the

scintillator edge and folded over it. Figure 5.13 shows a scan revealing a depressed

scintillator response.

Plastic Bags

In three instances sharp bumps were seen in one region of a section. One side would

show an increased response over severa} layers independent of the source position.

The other side had a de pression in the same region, also independent of the source

position. These towers wcre opened, and plastic bags were round where the bumps

were. These clear plastic bags had bcen used in the packing of the WLS bars when

they were shipped. They had been left on during the mounting of the WLS. The

plastic reflected the scintillator light back into sCÎntillator. This accounts for the

,

t

CHArTER.5. IlESlJLTS OF 6t)CO SCANS 79

incrt'a1>cd r(,~pO/1~e out the si de opposite the bag. They were easily removed.

5.3.5 Inhomogeneities

By fa.r tht' larg{'~t proportion of 'abnormal' 6OCO scans {ails into the category of bad

hornog('Ilt'lt) TI)('~(' wcre gC'nerally characteri2ed by bumps and dips which could not

Le explained dftcr a \'i~ual inspectIon of the tower. Sorne of these bumps or dips

OCCIII recl fur a group of scintIllators at one end of a stack. These were called steps.

Ot/wrs exllIlllu·d a ~l()pc in the plot ov('r the entire section. These were called slopes.

Slupps O('('\lfI illg al the edge of a HAC section were call('d shoulders. They did not

'i{'('rn tn 1)(' causcd by a WLS pl'Oblern because both sides of the tower were affected.

Fol' some, who have b\lmps and dips independent of source or WLS position, a

lcgiral connection is the composition of the scintillator themselves. For exarnple,

ail ('{fort had been made during the stacking to match scintillators from the same

ch(,IlIical bat( h witllln a tower. Perhaps for sorne towers, scintillator tiles from two

batchl'!>, 0/1(' of whlch wlth inferior optical qualities, were used, resulting in a step.

Another possibility is a grouping of scintillators with a higher or lower thickness than

a\'('rage This would affect the volume of scintillator exposed to the source and change

~illlIlarly IlBIIm'I' the response. Figure 5.14 shows an exarnple of a HAC l':.ection with

large un(,xlJlained inhomogeneities.

Others had bumps and dips strongly visible only with the source on the sicle

opposite the readout. This could be the result of faulty scintillators wlth a low

atlenuation length. Or it cauld be that the Tyvek paper somehow became clamp and

is stuck to the scintillator tile. This would destroy the total internaI reflection and

app('al' as 3. dccrease in the att.enuation length.

Tht' fru~t ratillg part is that none of these hypotheses can be reliably verified

withaul di'l1'cmtling a module. This was not an available option, as the stacking

machinc was not present at DESY. Even had it h"pn present, safety considerations

and time limitations would not have allowed th\..· \.llbmantling of a module. This also

makes it harder to est.imate the effect of su ch a fault on ,1 ,"al shower.

o'~

01

oœ

CHAPTER 5. RESULTS OF 6IJCO SCANS 80

,.l,~", 03

02

\ 0'

_ .... cd",..n.~co ..... ''''To •• ''.[ .. C2 L ,. .. - HAClI'

Figure 5.14: Bad homogenetiy

5.4 Scintillator Attenuation Lengths

By comparing the signaIs taken by two runs from opposite sicles of a tower a crude

estimate of the scintillator attenuation length can be made[24]. We can use a simple

model where the signal height is the produd of the responses from the WLS and from

the scintillator edge. The response from the scintillator is also attenuatcd. Wc take

two runs, 1 and 2, taken from the left and right sicles of the module respectively. The

WLS response is WR and WL . SR and SL are the scintillator edge responses. dis the

difference between the distances from the point of light production to the tWQ edges.

(

{

CHAPl'ER 5. RESULTS OF 6OCO SCANS

"'.0" 01720 AtlS 0 n~n-02

12 r-28

20

&

~t'~~06~.~U~~~~~~~O~1~~O~7~2~~O'7. Figure 5.15: Dlstnbution of the quantity e- d/>" in FCAL HACl ti/es

Then the atlenuation can be cxpressed as

-dl.\ e =

or

À = -d/21n E2LE1R.

E2RE1L

81

Using d = 16cm from the Monte Carlo, wc get 40 cm as an average attenuation

length for the FCAL HACI sections. As a comparison, beam tests conducted at

CERN arrived at an attenuation length of about 70 cm [191.

The difTerence between the attenuation lengths calculated above and those ob

tained during test beam runs can be explained by edge effects. The scintillator light

is produced very close to the edge. If there exists a gap between the Tedlar sleevf.:

and the edge of the scintillator, it would be possible for light which would have no."

mally bef'n ahsorbed by the black sleeve to escape and enter the WLS. The second

possibility is thal light which would have been absorbed by one of the two fluors and

then remitted isotropically goes instead directly out of the scintillator into the WLS.

CHAPTER 5 RESl'LTS OF GOGO SGANS

2

o .L........4-+ 1 II! 1 t 1. 1 I--.....~ ! 1 ... l,ft IlL

o 100 '00 600 8()Q 1 000 , 200 "00 1 600 'ourc. PO.I~'Of'\ ,""")

82

Figure 5 16: C(l'enkov llght seen zn a source scan wlth the $cmldlalors covered

The enhancement of signal from edge effects wh en the source is on the saille si de as

the rcadout makes the attenuation length sccm shorler.

Aithough the attenuation Iengths ca\culated from the 6OCO scans are different

from the beam test \'aInes, they can be used as a comparison with future 6OCo s,ans

to monitor changes to the attenuation length. The narrowncss of the di~tribution in

Figure 5.15, of about 1%, gives encouragement that the changt's of the same order

can be detected in the future. A change of 1 % in the value e- d/>' corresponds ta a

change of about 3% in .t

5.5 WLS Attenuation

The attenuation Icngth of the WLS can be determined by exarnining the amount

of Cerenkov light emitted iL the sections of the WLS not facing the scintillator(31).

In one of the special measurements a strap wa.s removed and the edges of ail the

scintillalors of a tower were covcred. Only the Ccrenkov light in the WLS r€'achcd

the PM. Figure 5.16 shows the signal when the scan was taken. The gentie slope

{

CHAPTER 5. RESULTS OF MJCO SCANS 83

III

1 .!l 'ni .... n. ~ ...... 1101

l1li5 2!l2 1

" P .0

3!l

JO

2!l

20

l!l

'O

!l

~OOO Mm

Figure 5.17: Attenuation lengths of EMC WLS

as the source moves away from the end of the WLS is due to the attenuation of the

WLS.

This cannot be done after the modules are finally installed in the ZEUS detector.

It is difficult to rernove the contribution {rom the scintillators to the signal and leave

only the Cerenkov light. Instead the regions where the WLS merely transports light

and does not see scintillator light must be used.

For EMC WLS, this means fitting an exponential to the signal along the HAC

sections. For the HACI section an exponential is fitted along the HAC2 section. The

attenuation length of the HAC2 WLS cannot be measured in this way. The function

fittcd to the inactive regions is

where L is the length of the WLSj x is the distance to the to PM; r is the reflection

coefficient of the end reflector. The distribution of attenuation lengths for FCAL

EMC WLS is shown in Figure 5.17.

Chapter 6

Conclusions and Outlook

6.1 Scanning in the ZEUS Hall

Although the outside scanning method pro\'ed to be (. feliabl<, way ta scan ail the

modules, it unfortunately cannot be used whil{' the modu\<,s arp lIlstalll'd in Ill(' ZEUS

detcctor. Preparations are being made to implOvc the lIl,>ide ..,Cêlllllillg JlwtllOd f,() il

can be reliably used for at least a few modules. First, a ncw source is \wÎng prC'pared

which will use a single long stpel tube Ct ntaining almost the Plltile }ellgth of the piallo

wire inside. This will avoid the bunching up of the \Vile which c,1ll~('d the fli<tion

plOblems \Vith the C'ddier ~Olll'C(,S. S~condly, mo!)t d the bra,,:, gllldp t\l\J<''' in IH'AL

plus those in the special half FCAL modules have bcell chcckcd fOI hadly glued joinls

with a Jummy source wire pushed in by hand. 14 brass tubes hac! joillt'> wllich ù!wlwd

during testing. In a further 66 brass tubes the end of the WLS (,,~~et 1(' coulcl not

be reached. A few of the tubes were repairC'd. Most of thC' 1C',>t hrld tlieir ('lltfan(('~

closed. It is envisioned that a new outside sc,wner wlll be built which will flln alollg

the top C-arm of the special modules to scan the towC'rs dll'f"_ tly il \)0\1(' the !WdJlJ

pipe in FCAL and RCAL while the clam shell is opf'ned. Although only Ol\(' sirle of

the towers can be scanned in this way, it is hopcd Lhat cnough illfO! rtIatioll (an be

obtained to decide whether inside scanning is warranted.

84

CIlAPTER 6. CONCLUSIONS A.ND OUTLOOI\ 85

6.2 Conclusions

6°['0 hùlll(t~ scans ha\ e becn conduct.ed from the inside for sorne and outside for

ail FeAL and ReAL modules. The source scans have proved t.o be very useful in

df·tecting simple assembly faults. Most of the faults detected involving shadowed

scÎntillators and faulty WLS have been corrected. Large inhomogeneities are often

sœl\ and have no confirmed cause, but are speculated ta be connected with the

sciutillatùr uniformity. It is not clear what effcds these illhornogeneitics will have on

the calotim('ter's performance. A surnmary of the 'fault-.' ! ,111 be found in TaLle 6.l

and Table 6.2. Although the cobalt scaus cannat give tllL' true attenuation lengths

of tll(, o.,Cilltill,ltors clue to eJge eff,-"cts, it may be possible to monitor changes larger

titan a ft·,,· percent to the attcllllation lengths. The WLS attenuation lengths can also

b(' rca::,oll<\bly monitOlcd, though only for the rcgions rcceiving no direct scintillator

light. A ncw and more mechanically rcliable SOUfce is being prepared to continue

inside scanning on the installed modules. Due ta mechanical problems with the

inside guide tubes, only sclectcd towers are intended to be rescanned in the future.

CH.-1PTER 6. CONCLUSIONS AND OUTLOO/( 86

1

Section

Fault EMC/IIACO HACI IIAC2 -- c--'

Bad Stacking 3.0% 80% 6.5% --

Shiftcd WLS 1.3% 7.6% 3.7% ----

Shado\'v'ed Scint. 0.5% 0,2% 0.9% --

Bad Homogeneity • 0.5% a.3% 0..1% --

Shoulder • - 1.1% :1.:J% --t-------

Stcp· 0.1% 11..5% !J. J (Ic, --

Slope • 2.8% ~r:o/: -r--~~~~j BumpjDip • - 4.6%

-

listing contains on1y faults ~ 7%

• no satisfactory explanation yet found

measuf<>d towers: 460

of that 264 EMC towers

of that 196 HACO towers

Table 6.1: Summary of Faults in FeAL

J

l

t

CHAPTER 6. CONCLVSIO.VS AND OT..:TLOOJ(

Section

Fault EMC/HACO HACI

Bad Stacking 1.4% .5.8%

Shifted WLS 0,4% 1.1 % ---

Shadowed Scint. - 1.1%

Bad lIomogeneit.y • 2.5% 13.3%

Shoulder * 2.1% 1.3%

Step • 4.7% 12.2%

Slope • 1.0% 1') 'Je;;( _._ 0

BumpjDip • 2.0% 20.1%

listing cantains anly faults ~ 7%

• no satisfactory explanation yet found

measured towcrs: 452

of that 256 EMC towers

of that 196 JlACO towers

Table 6.2: Summary of Fau/ts in ReA L

8ï

(

(

Bibliography

[I] M. GelI-Mann; A schematic model of Baryons and Mesons; Phys Lett. 8. (1964) 214-215

[2] E.D. B100m et al.; High Energy Inelastic e-p Scattering at 6° and 10°; Phys. Rev. Lett. 23 (1969) 930

[3J G. Wolf; HERA: Physics, Machine al1d Experiments; DESY 86-089 (1986)

[4J Proceedings of the study of an ep facility for Europej Ed. U. Arnaldi; DESY 79/48 (1~79) 391-394

[5] Proceedings of the liERA Workshop Vol. 1 an cl 2; Ed. R.D. Peccei; DESY (1988)

[6] ZEUS Collaboration; The ZEUS Detector, Statu& Report 1989 (1989)

[7] J .E. Brau and T.A. Gabriel; Nuel. Jnstr. and Meth. A238 (1985) 489

[8] R. Wigmans; On the Energy Resolution of Uranium and other Hadronic Calorimetersj Nucl. Instr. and Meth. A259 (1987) 389

[9] H. Brückmann et, a1.; Hadron Sampling Calorimetry, A Puzzle of Physicsj Nucl. Instr. and Meth. A263 (1988) 136

[10] M.J. Berger and S.M. Seltzerj Tabl~s of Energy Losses and Ranges of Electrons and Positrons; NASA-SP-3012 (1964)

[U) U. Amaldi; Fluctuations in Calorirnetry Measurementsj Phys. Scripta 23 (1989) 409

[12] Review of Parti cie Propertiesj Phys. Lett. 1708 (1986) 45

[13] E. Ros; ZEUS-Note 86-135 (1986) (unpublished)

[14] R. FCl'llo""j Introduction to experirncntal particle physicsj Cambridge University Press (1986)

[15] M. Catallesi et al; Nucl. Jnstr. and Meth. A260 (1987) 43

[16] A. Sessoms et al.; The segmented calorimeter: A study of hadron shower strucLure, Nucl. lnstr. and Meth. A195 (1982) 461

88

, BIBLIOCUAPHY 89

[17] H. Abramowicz et al.; Nucl. Instr. and Meth. 180 (1981) ·129

[18] R. Klanner; Test Program for the ZEUS Calorimeter; DES\" 8;·058 (1~)8Î)

[19] ZEUS CalorimetC'f Group;Comtruct ion and B<'am Test of t.he ZEl1S Forwaro and Rear Calorimcter; to be puhlished in N.I M.

[20] R. Blankers et aL; Const! uetion of the ZEUS ForwardjRt'ar Calorinwter Modules at NI KEF; NIKHF..F-H 90-11 (unpublished)

[21] J. Straverj PhO Thesisj NIKHEF (1990)

[22) R. Carwinj The design of liquid scintillator cells, Rev. Sei. Justr. 23 (1952) 755

[23] 1. Gialas et aL; Effcct of light output nonuniformity on the ZEUS barrel calorimeter calibration; ZEeS-Note 89-109 (unpublished)

[24) U. Behrens et aL; First Measurements with Hadioé\ctive Sources III the FCAL Prototype Modules; ZEUS-Note 88-56 (unpllbli~hed)

[25] R. Krüger; Diplom A'rbeit; Universitat Hamburg (1990)

[26] D. Gilkinson; private communication

[27] R. Brun and J. Zo11; ZEBRA user guide, CERN program library t~ntry QIOO; CERN, Geneva (1987)

[28] W.R. Nelson, H. JIirayama, and 0.\\'.0. Rogers; SLAC Report 165, Stanford: SLAC (1985)

[29J G. Gloth; Oiplom Arbeitj Univcrsit~it Hamburg (1991)

[30] G. Stairs; private communication

[31) F. Meyer; Oiplom Arbeit; UniversiUit Hamburg (1991)

The 60Co Calibration of the ZEUS Calorimetersdigitool.library.mcgill.ca/thesisfile61196.pdf · The 60Co Calibration of the ZEUS Calorimeters by Ling ... 3.1 Radzation Dose from a

Documents